Home

Active Control of Feedrate and Spindle Speed to Extend Tool Life

Contents

1. L Spindle Speed rpm Figure 21 Spindle Speed Output Fuzzy Set T T T T T T T N fN fZ fP fBP 1400 1500 1600 1700 1800 1900 Feedrate mm m Figure 22 Feedrate Output Fuzzy Set Figure 19 shows that the spindle speed set ranges from 1500 rpm to 3500 rpm centering on 2500 rpm The spindle negative sN and spindle positive sP both have a range of 840 rpm with the peak shifted 20 rpm towards the center of the set The two outer functions are standard have a range of 500 rpm their peaks at the borders of the sets The zero function was given a significantly larger range than the other sets total range being 1400 rpm with the peak right at the center of both the function and the set Figure 20 shows that the feedrate set ranges from 1300 mm m to 2300 mm m centering on 1800mm m The two outer functions again have their peaks on the borders of the set this time with a range of 250 mm m each The negative and positive sets were given a range of 500 mm m with the peaks shifted 50 mm m towards the outside of the set Again the zero function was given a much larger range than the rest with a range of 800 mm m and the peak both in the center of both the function and sets range The actual results of running through a simulation with the fuzzy set are displayed below in Figure 21 Fuzzy Logic Control 700 680 660 640 620 600 Force N 580 560 540 520 500 0 0 5 1 1 5 2 2 5 3 3 5
2. else there is a controller so enable the start button else startButton gt setEnabled true clearOldData theController set theController to the one from the ParamTable theController getAtIndex the Ptable gt currentItem else startButton gt setEnabled false toolComboBox gt currentItem will check if the current tool op combo has a controller opComboBox gt currentItem opComboBox opComboBox gt currentItem opComboBox 0 0 Main Program include lt qapplication h gt include forml h include qwt plot h int main int argc char argv QApplication a argc argv Forml w w show a connect amp a SIGNAL lastWindowClosed amp a SLOT quit return a exec
3. currTool printf Starting Timer n tehtimer startTimer int 10 take in data impliment control algorithm communicate with NC plots void Forml timerEvent QTimerEvent flash the alarm if it is on if alarmFlag 1 alarmCounter alarmFrame gt show if alarmCounter gt 20 if alarmCounter gt 40 alarmCounter 0 change to red alarmFrame gt setPaletteBackgroundColor Qt red alarmFrame gt repaint true alarmLabel2 gt setPaletteBackgroundColor Qt red alarmLabel2 gt repaint true else alarmFrame gt setPaletteBackgroundColor Ot black alarmFrame gt repaint true alarmLabel2 gt setPaletteBackgroundColor Qt black alarmLabel2 gt repaint true Alarm wasnt currently acitvated so proceed else if too high of force Alarm if sqrt getCurrForceSquared theDataCollector gt currTool gt maxAllowableForce ALARM stop the timer killTimer tehtimer perform the alarmRoutine aka spindle off back off the piece alarmGUI now display the alarm in GUI alarmFrame gt show alarmFrame gt showFullScreen alarm not active and the force is acceptable so store gui and control else because of the 10ms alarm 100ms control dataCounter if dataCounter gt dataCounterTop get the ADC force values getAndStoreForces theDataCollector printf GOT AND STORED FORCES n this might not
4. Fix return NULL Queue empty check to make sure we will not look before the beg of the queue else if which queue gt tail which queue gt beg void ret which queue gt end return ret tail peek else which _queue gt tail void ret which queue gt tail which queue gt tail return ret int isFull Queue which queue compare max cells to cells used if which_queue gt cells_ used gt which_queue gt max_cells return true return 1 else return false return 0 ifdef cplusplus extern C fendif ifndef TOOL H define TOOL H Struct to define a tool and related functions El typedef struct tool char toolName The name of the tool float maxAllowableForce The maximum allowable cutting force for the tool unsigned short number for easy ID Tool Tool brief Creates a tool by allocating mem for the struct and its members and initializes it param t float the Controller gain of the new Controller param mat float The controller signal param maxForce float Gain used when adjusting the feedrate param refForce float The previous signal sent by the controller return Controller returns a pointer to the newly allocated Controller Tool create Tool char name float maxForce unsigned short num TENG KR fendif ifdef _ cplusplus tendif include lt stdlib h gt include lt string h gt in
5. Active Control of Feedrate and Spindle Speed to Extend Tool Life During CNC Milling Processes A Major Qualifying Project Submitted to the Faculty of WORCESTER POLYTECHNIC INSTITUTE In partial fulfillment of the requirements for the Degree of Bachelor of Science by Joseph Driscoll RBE Brenden Gibbons RBE In Collaboration with Huazhong University of Science and Technology Partners Yuan Liu and Yuling Niu Date 14 December 2011 Approved Professor Yiming Kevin Rong Major Advisor ME Abstract This project involved design and partial implementation of an active tool wear monitoring and prevention platform In order to assess the sponsor company s needs a requirements analysis was performed After researching and analyzing the current state of the art a program was written in C and C to actively monitor axial cutting forces and control spindle speed and feedrate of a CNC to maximize tool life through various control methods In order to complete the system a comprehensive user friendly GUI was created Acknowledgements We would first like to thank HUST for granting us the use of their facilities and their hospitality We would also like to thank Professor Rong Professor He Professor Li Dr Hongqi and Zhigang Yang for providing support in both setting up and executing the project Additionally our thanks go out to Huazhong Numerical Control our host company Finally we would like to thank our partn
6. en inc inc tinc tinc tinc int dif lude lt stdio h gt lude lt string h gt lude lt errno h gt lude lt stdlib h gt lude queue h outputFloatLine FILE stream float f if the inputted file stream hasnt been terminated LE OA stream 0 print to file the float fprintf stream 2f n f file pointer is terminated or null else print error printf Error d s n errno strerror errno return 0 return 1 int outputFloatTab FILE stream float f if the inputted file stream hasnt been terminated if stream 0 print to file the float fprintf stream S 2f t El file pointer is terminated or null else print error printf Error d s n errno strerror errno return 0 return 1 int outputStringTab FILE stream char string if the inputted file stream hasnt been terminated if stream 0 print to file the string fprintf stream Ss t string file pointer is terminated or null else print error printf Error d s n errno strerror errno return 0 return 1 int outputStringLine FILE stream char string if the inputted file stream hasnt been terminated if stream 0 print to file the string fprintf stream Ss n string file pointer is terminated or null else print error printf Error d s n errno strerror errno return 0
7. which is a percentage of ideal feed rate and return it return FC gt idealFeedRate FC gt currentFeedRate takes in array of input truths and applies fuzzy logic rules and defuzzifies storing the result in the fuzzyController void applyFuzzyLogic fuzzyController FC float errorTruths float deltaErrorTruths sums of Areas and sums of Areas Centroids float FACSum 0 float FASum 0 float SACSum 0 float SASum 0 loop counters int i 3 running through error for i 0 i lt 5 i running through change in error for j 0 3 lt 5 j if there is a triangle dont use 0 because of floating point errors if deltaErrorTruths j gt 01 amp amp errorTruths i gt 01 defuzzification sum the area centroid of all the triangles with truth then divide by the sum of the areas FACSum getAreaOfMember FC gt feedOutputMFD fuzzAnd deltaErrorTruths j errorTruths i FC gt feedTable j i getCentroidOfMember FC gt feedOutputMFD FC gt feedTable j i FASum getAreaOfMember FC gt feedOutputMFD fuzzAnd deltaErrorTruths j errorTruths i FC gt feedTable j il SACSum getAreaOfMember FC gt spindleOutputMFD fuzzAnd deltaErrorTruths jl errorTruths i FC gt spindleTable j i getCentroidOfMember FC gt spindleOutputMFD FC gt spindleTable j i SASum getAreaOfMember FC gt spindleOutputMFD fuzzAnd deltaErrorTruths jl error
8. www personal umich edu ykoren uploads Adaptive control systems for machining pdf gt Liang Ming Tet Yeap Saeed Rahmati and Zhixin Han Fuzzy Control of Spindle Power in End Milling Processes International Journal of Machine Tools and Manufacture 21 14 2002 1487 496 Science Direct 12 Oct 2002 Web Autumn 2011 lt http www sciencedirect com science article pii S0890695502001402 gt Oh Young T Gi D Kim and Chong N Chu Design of a Drilling Torque Controller for a Machining Center The International Journal of Advanced Manufacturing Technology 22 5 6 2003 329 35 Springer Link Web Autumn 2011 lt http www springerlink com content tag7gj1qdppb8wyn fulltext pdf gt PCM 3355 User Manual Advantech Co Ltd lt http support advantech com tw support DownloadDatasheet aspx gt Kistler Type 5070A DataSheet 2010 The Kistler Group lt http www kistler com mediaaccess 5070A BP _000 485e 03 10 pdf gt Operating Instructions Multichannel Charge Amplifier 1996 The Kistler Group lt http ust fme vutbr cz obrabeni rozvoj Kistler CNC 5019B Description En pdf gt QTimer Class Reference 2011 Nokia Corporation lt http developer qt nokia com doc qt 4 8 qtimer html gt Appendix A Full UML Diagram MazyClonrdla createFuzzy Control enya fuzzyContrdiar SetaulC natt uy Cora float float float float float vdd getTrutha mF D f
9. malloc sizeof float fp3 threer Force enqueue the forces into their appropriate queues printf Forceenqueuing Dereferenced ONE f n fpl forceEnqueue the dataCollector gt xForceQueue void fpl printf Forceenqueuing Dereferenced TWO f n fp2 forceEnqueue the dataCollector gt yForceQueue void fp2 printf Forceenqueuing Dereferenced THREE f n fp3 forceEnqueue the dataCollector gt zForceQueue void fp3 return 0 float forceFloat dataCollector the dataCollector make some float pointers float totalForce float xp float yp float zp look at the most recent queue entries xp float peekTail the dataCollector gt xForceQueue yp float peekTail the dataCollector gt yForceQueue zp float peekTail the dataCollector gt zForceQueue magnitude the forces totalForce xp xp yp yp zp zp return totalForce float getCurrForceSquared dataCollector the dataCollector float totalForce float xp toForce pullAnalogPort the dataCollector 0 the dataCollector gt mode float yp toForce pullAnalogPort the dataCollector 1 the dataCollector gt mode float zp toForce pullAnalogPort the dataCollector 2 the dataCollector gt mode totalForce xp xp yp yp zp zp return totalForce float getForceFromFS dataCollector the dataCollector Tool t cuttingOperation co Controlle
10. new fuzzyController gt idealSpindleSpeed 0 new fuzzyController gt idealFeedRate 0 new fuzzyController gt idealCuttingForce 0 new fuzzyController gt maxForce 99999 return the fuzzyController pointer return new fuzzyController fuzzyController defaultCreateFuzzyController float eNeg DE float dePos float recFeed float recSpindle MAKE THE FUZZY CONTROLLER memberFunction mfl memberFunction mf2 memberFunction mf3 memberFunction mf4 memberFunction mf5 mFD mFD1 mFD mFD2 mFD mFD3 mFD mFD4 input error float step ePos E 3 mfl createMemberFunction E 3 step E step eNeg float E float ePos float deNeg float default membership functions equilaterally spaced mf2 createMemberFunction mf3 createMemberFunction mf4 createMemberFunction E E 2 step E step mf5 createMemberFunction E step E 3 step ePos mFD1 createMFD 0 mfl mf2 mf3 mf4 mf5 input delta error step dePos DE 3 mfl createMemberFunction DE 3 step DE step deNeg mf2 createMemberFunction DE 2 step DE DE step E 2 step E E step E step E step E mf3 createMemberFunction DE step DE step DE mf4 createMemberFunction DE DE 2 step DE step mf5 createMemberFunction DE step DE 3 step dePos mFD2 createMFD 0 mfl mf2 mf3 mf4 mf5 output spinlde mfl createMemberFunction 15 51 33 mf2 createMemberFunction 48 84 66 mf3 createMemberFunction 75 1 25 1 mf4 c
11. ze 04 gt 0 6 i i boob AA A AAA A OL LL 50 40 30 20 10 0 50 40 30 20 10 0 50 40 30 20 10 0 Time s Time s Time s Cutting Force N Spindle Speed rpm Feed Rate mm min Select Tool Select Operation Basic Advanced 3 3 ballendmil Y 30 Contouring aluminum o s Reset Quit Figure 24 Final GUI Design The foremost specified requirement of the GUI was to display real time plots of the cutting force spindle speed and feedrate This was accomplished by expanding upon the qwtPlot widget from the QWT library Each plot dynamically changes its y axis scaling to best fit the range of incoming data The x axes range from zero to a negative number of seconds that is specified in pre processor defines of the program The three graphs are very large and placed prominently at the top of the GUI Display boxes below each of the three plots show the numerical value of the current data They are updated along with the three plots every 100 milliseconds They cannot be clicked on or typed in and are intended to be useful when the data range in a plot is large enough to create confusion and misestimating upon visual inspection On the lower left hand side of the GUI drop down boxes for selecting both a tool and an operation are present Although initially empty they can be added to by using features in the Advanced Tab When To the lower right of the GUI there are two tabs Basic and Advanced All
12. 1 0 5 10 15 20 25 30 35 40 45 50 Time 15 Figure 19 Synchronous PID Control Response After some tuning the final gain values for the system were Table 5 Synchronous PID Gain Values Feedrate Gains Spindle Speed Gains Kp 1 Kp 1 5 K 11 K 15 Kp 01 Kp 03 These values result in a very rapid adjustment to the desired force while overshooting the desired force by a small amount In more specific terms a rise time of about 35s a settling time of about 9s and about an 11 overshoot Figure 18 below displays the tuned response of the asynchronous PID controller Force 750 700 650 600 Force N 550 500 0 5 10 15 20 25 30 35 40 45 50 Time 15 Figure 20 Asynchronous PID Control Response The gains used for this system were as follows Table 6 Asynchronous PID Control Gain Values Feedrate Gains Spindle Speed Gains Kp 1 Kp 1 3 K 01 K 03 Kp 05 Kp 01 These values unlike the synchronous version provide a much slower rise with a very small percent overshoot Because the rise time of the system is so slow the rise and settling times are about the same at roughly 8s with about a 3 overshoot While the settling time is actually smaller than the synchronous function the asynchronous function takes a very long time to reach the actual desired value 4 1 3 Fuzzy Logic Control System Development Tuning fuzzy controller resulted in the two following output sets
13. 4 4 5 5 Time 1s Figure 23 Fuzzy Logic Control Response Because of the zero values for the two output sets I 800mm m for feedrate and 2500 rpm for spindle speed the desired force came out to be 507 28N instead of the 500N that was used for the PID simulation Because Simulink runs in discrete time steps it is difficult to determine any values such as rise and settling time Though the fuzzy the simulation results show a correction time of 4s as the first 1s involved getting the simulation to run properly as a result of the Simulink setup 4 2 Software System Development This section describes what was achieved in terms of the overall software codebase and the Graphical User Interface 4 2 1 Final Graphical User Interface Design Figure 22 below shows the multi functional GUI As the guidelines for the user interface were not outlined in depth by HNC the team tried to create one that would encompass a range of usage and abilities from simple to complex This would enable basic users i e factory workers to use the program simply to monitor cutting operations while advanced users could create and save individual control systems that best fit specific cutting tool and operation combinations in Tool Wear Monitoring and Control x Cutting Force Spindle Speed Feed Rate 0 6 i F y i i 0 6 i i i D t i i wa 0 24 5 0 Tr a zZ 0 24
14. Control Algorithms A e o o o 30 4 1 2 Proportional Integral Derivative Control System Development oooooonnccicccncccoccnoncnnnonos 30 4 1 3 Fuzzy Logic Control System Development srnronvrrnnvrrrrnrrnrrrnnrrrnrrrrrnerrnrensrernrerrrneenn 32 4 2 Software System Development aa4v42v2429ss NG 34 4 2 1 Final Graphical User Interface Design rrnrrnnrrrnrrrnvrrnnrrrernrrrnrrnnrrsnnrrrrnrrrneennessnnsrrrsernn 34 FA Data Collection Results ici Sc NNN 40 4 5 CNC Communication Results vic dicta 42 A sos 42 E uer Seed 43 6 F t re Works ass STEG 44 Gil Response Ume iiss rian p Aa EE EA AR ERE A a Naah 44 6 2 Communication with the Numerical Controller soranvrennrnrnrrrrnrrrnvrrrnrnrrrnrrrnernrerrnrerernernn 44 BJ TINE TE MNA 44 64 Database Implementations sicccstuscssacsaieic scans ase tek la dots tauleiceatiseds cites hesibedas a tueiviteanke d 44 6 Load Band Momitoting E s 44 Works Ude ee 46 Appendix A Full UML Diagram ti 47 Appendix Bi Matlab Test Code vassere ae te gdenaieiaess 48 Appendix Ge Program o e o RAA 50 Table of Figures Figure 1 Large Generator SHALE ii dr 8 Figure 2 Top Original Control S k caida 9 Figure 3 Milling Machine Table and Spindle cece ceeessceseeeeeeeeceaecenecaeeseeeeeesseeaecaeeneeens 11 Figure 4 Milling Machine Computer Numerical Controller erronrronororernrnonvnnnrenrrerersrernrnnnee 11 Figure 5 PID Control Loop Di ri ns 13 Figure 6 Fuzzy Logic Control Loop
15. DRV_GetErrorMessage ret err msg printf err msg s n err msg DRV_DeviceClose amp fd return 1 return voltage sim Ew A we must create a simulated AD value that is reasonable this should give a lowish standard deviation int result makerand 128 makerand 128 makerand 128 makerand 128 the_dataCollector gt lastvalue the dataCollector gt lastvalue result return float result float toForce int rawInput simply convert raw ad data to force the raw output voltage rangle of the charge amp is between 10 and 10V there is a 12 bit AD converter so between 0 and 4096 resolution first multiply by 0048828125Volts Digital to convert to a voltage float result result rawInput 0048828125 Next subtract ten volts to get the proper range result result 10 finally multiply by Newtons Volts result result 10000 20 return result int getAndStoreForces dataCollector the dataCollector get the forces float oner toForce pullAnalogPort the dataCollector 1 the dataCollector gt mode float twor toForce pullAnalogPort the dataCollector 2 the dataCollector gt mode float threer toForce pullAnalogPort the dataCollector 3 the dataCollector gt mod allocate some memory float fol fpl float malloc sizeof float fpl oner float fp2 fp2 float malloc sizeof float 7 Di e fp2 twor float fp3 fp3 float
16. and breakage monitor is already in use by the customers of Huazhong Numerical Control Co Ltd that machine the piece above however this has become obsolete Running on DOS and implementing outdated signal processing methods this system needs to be replaced To create a new and effective control system a list of objectives was created and they are as follows detect machine tool abrasion and breakage through the analysis of cutting forces extend tool life by actively controlling feedrate and spindle speed with a response time of less than 100ms continuously display plots of the spindle speed feedrate and cutting force develop and efficient and stable signal processing method and a user friendly interface back the tool off of the workpiece within 10ms of extreme tool abrasion or tool breakage and finally display an alarm if significant tool wear or tool breakage has occurred 2 Background The Background Chapter will convey general background information concerning the host company its competitors and iterations of the specific product that this project focuses on It will also provide a brief literature review of similar past works and related topics 2 1 Company Background Huazhong Numerical Control Co Ltd HNC was founded in 1994 in Wuhan China as a researcher developer and producer of Computer Numerical Control CNC systems Since its inception HNC has become one of the leading CNC manufacturers in the world offering i
17. be right but calculate the cutting force adam method currForce forceFloat theDataCollector currForce sqrt currForce printf Current Force f n currForce Control multiControl theController currForce 1 update the forceQueue float fpl fpl float malloc sizeof float fpl currForce forceEnqueue forceQueue void fpl forceShow gt setText QString number currForce f 3 qwtShiftArray force dataSize 1 force 0 currForce forcePlot gt replot update spindleSpeed and store in queue float fp2 fp2 float malloc sizeof float fp2 theController gt currSpindleSpeed fp2 spindleControlFunction theController currForce forceEnqueue spindleQueue void fp2 spindleShow gt setText QString number fp2 f 3 qwtShiftArray spindle dataSize 1 spindle 0 fp2 spindlePlot gt replot update spindleSpeed and store in queue flost fp3 fp3 float malloc sizeof float fp3 theController gt currFeedRate fp3 feedControlFunction theController currForce forceEnqueue feedQueue void fp3 feedShow gt setText QString number fp3 f 3 qwtShiftArray feed dataSize 1 feed 0 fp3 feedPlot gt replot CONTROL free the memory free fpl float a 600 make the static int zeta 0 if zeta 100 zeta 0 qwtShiftArray y 600 1 y 0 zeta forcePlot gt replot
18. f The optimal spindle speed for the current operatio The optimal cutting force for the current operation brief Creates a Controller by allocating mem for the struct and its members and initializes loat the Controller gain of the new Controller float The controller signal loat Gain used when adjusting the feedrate float The previous signal sent by the controller loat The current feed rate of the NC float The optimal spindle speed for the current operation float The optimal cutting force for the current operation pointer to the tool used for the current operation Has a maxForce ller returns a pointer to the newly allocated Controller Controller create Controller void init as Fuzzy Controller cont float eNeg float E float ePos float deNeg float DE float dePos float recFeed float recSpindle float Ks float u float doc void init_as_asynchPID Controller cont float desF float fkp float fki float fkd float skp float ski float skd float errbound void init as synchPID Controller cont float desF float fkp float fki float fkd float skp float ski float skd will use control function of whatever type of controller void multiControl Controller cont float currForce double timestep brief Control function for determing the correct feed rate param cont Controller A pointer to the controller in question param currForce float The current cutting
19. feedrate of the machine 3 2 6 Control Algorithms Overview Sections 2 6 1 and 2 6 2 provide brief overviews of the two control methods used in the project 2 6 1 Proportional Integral Derivative Control A PID controller is a basic control algorithm that calculates error by taking the difference between the current state and desired state A basic control block diagram can be seen below in Figure 5 To calculate the proportional term the error is multiplied by some gain Kp The integral value is calculated by summing the error with all previous errors and multiplying the new value by some gain Ky To determine the derivative value the change in error over time is calculated by subtracting the previous error from the current error and dividing by the change in time since the two measurements This value is then multiplied by some gain Kp All gain values are determined through experimentation The three terms are then summed together resulting in an output value corresponding to necessary change The process then loops until the object being controlled has reached the desired state Desired State _ e t T Control Signal _ Kr S elt O m fy d Feedback Signal Measured State Figure 5 PID Control Loop Diagram 2 6 2 Fuzzy Logic Control Fuzzy logic controllers are much more complex and involved than a PID controller Though the basic forms of the control block diagram seen below in Figure 6 and above in Figure 5 for e
20. force would be based on the output and then feed that force back in The equation used to accomplish this is as follows Equation 1 Cutting Force Calculation V U F Kkexa x Where F is the cutting force in Newtons a is the depth of cut in mm V is the feedrate in rpm p is the number of teeth on the tool N is the spindle speed in millimeters per minute mm m Ks is the specific cutting force in Newtons and u is some value ranging from 6 lt u lt 1 Both Ks and u depend on the tool and workpiece material 1 The values chosen for testing were selected as average values based on knowledge of machining and industry standards The values used were as follows Table 1 Cutting Force Equation Historical Trends Ks 500 N a 4mm p 4 teeth u 8 with a desired cutting force of F 500 N Various starting forces were chosen both above and below the desired cutting force to verify that the algorithms work in multiple different scenarios 3 2 1 Proportional Integral Derivative Control Testing Tuning the two PID controllers involved adjusting the gain values corresponding to the feedrate and spindle speed until an acceptable response was produced This means having fast rise and settling times and a small percent overshoot This is the same process that needs to be done for every new cutting operation The code used to test the two controllers can be found in Appendix B 3 2 2 Fuzzy Logic Control Testing To test the fuzzy c
21. guards ifndef QUEUE H define QUEUE H Struct to define a queue each entry can hold a pointer to anything PJ typedef struct queue void tail Pointer to tail FI element of queue void head Pointer to head FO of queue void beg Pointer to the beginning of the queue void end Pointer to the end of the queue int max cells Maximum number of entries in the queue int cells_used Currently used number of cells Queue Nbrief Creates a queue by allocating a queue structure initializing it and allocating memory to hold the queue entries param max_cells int Maximum entries in the queue return Queue Pointer to newly allocated Queue structure NULL if error Kl Queue create queue int max cells brief Enqueues a pointer onto the tail of a Queue param which queue Queue Pointer to queue you want to enqueue onto param ptr void Pointer to be enqueued return int 0 if successful 1 if not y int enqueue Queue which_queue void ptr Forces a pointer onto the tail of a Queue regardless if it is full param which_queue Queue Pointer to queue you forceEnqueue param ptr void Pointer to be force enqueued return int 0 if successful 1 if not E int forceEnqueue Queue which_queue void ptr brief Dequeues a pointer from head of Queue and returns it param which_queue Queue Pointer to
22. mFD in question param i int Indicates which member function to get the area of Xreturn float The desired area 57 float getAreaOfMember mFD mFD float truth int i endif ifdef _ cplusplus endif include lt stdlib h gt include lt stdio h gt include mFD h mFD createMFD int i memberFunction BNeg memberFunction Neg memberFunction Zer memberFunction Pos memberFunction BPos pointer to the new mFD mFD new mFD allocate mem for mFD new mFD mFD malloc sizeof mFD if new mFD NULL return NULL Error unable to allocate set the member triangles new_mFD gt bigNegative BNeg new_mFD gt negative Neg new mFD gt zero Zer new mFD gt positive Pos new mFD gt bigPositive BPos return pointer to new MFD return new mFD void getTruths mFD mFD float input if the input is off the charts negative the big negative truth will be 1 if input lt mFD gt bigNegative gt left mFD gt truths 0 1 else treat normally else mFD gt truths 0 getTruth mFD gt bigNegative input middle three memberfunctions mFD gt truths l getTruth mFD gt negative input mFD gt truths 2 getTruth mFD gt zero input mFD gt truths 3 getTruth mFD gt positive input if the input is off the charts negative the big negative truth will be 1 if input gt mFD gt bigPositive gt right mFD gt truths 4 1 else treat normally els
23. of the buttons required for basic usage of the GUI exist under the Basic tab These buttons would be used by a typical factory worker when monitoring a cutting operation with previously defined parameters and controllers Figure 23 below shows the basic tab a Reset Figure 25 GUI Basic Tab The green Start button begins cutting force monitoring control of spindle speed and feedrate and real time data collection and plotting It is dynamically enabled when a valid exists for the selected Tool and Operation and disabled when no such controller exists The red Stop button temporarily suspends cutting force monitoring real time control of spindle speed and feedrate and data display Pressing Stop does not clear or reset any data The Reset button clears all existing data and the three plots It is used when changing the Tool and operation combination or when performing a new cut as all stored errors within the current controller will be cleared The Quit button to the absolute lower right of the interface clears all data tools operations and controllers and exits the program 4 2 1 1 Advanced Tab The Advanced Tab shown below in Figure 24 houses four buttons Basic Advanced Save Data New Tool New Controller New Operation Figure 26 GUI Advanced Tab The Save Data button writes all currently stored data to a text file for future viewing The file is
24. string to print return int 1 if success 0 if fail outputStringLine FILE stream char string brief write a Queue in a column to a file param file FILE a pointer to the file we wish to write to param queue Queue a pointer to the Queue we wish to write return int 1 for success 0 for failure Kl int writeQueueToFile FILE file Queue queue brief Creates a file with a given name and writes a queue to it param fname char the file name as a string param queue Queue a pointer to the Queue we wish to write return int 1 for success 0 for failure FU int writeQueueToFileName char fname Queue queue brief write 6 queues formatted in columns to a file param file FILE a pointer to the file we wish to write to param queuel Queue a pointer to the first Queue we wish to write param queue2 Queue a pointer to the second Queue we wish to write param queue3 Queue a pointer to the third Queue we wish to write param queue4 Queue a pointer to the fourth Queue we wish to write param queue5 Queue a pointer to the fifth Queue we wish to write param queue6 Queue a pointer to the sixth Queue we wish to write param time the time step return int 1 for success 0 for failure Kl int writeAllTheQueuesFormatted FILE file Queue queuel Queue queue2 Queue queue3 Queue queue4 Queue queue5 Queue queue6 double time fendif ifdef _ cplusplus
25. the GUI is shown in Figure 14 below hd Too Wear Monitoring and Control x Cutting Force Spindle Speed Feed Rate A Pony oo oor bf Mi po A fingre EE ech FE A 04 i E a 02 0 i o i le EN eect 0 2 0 4 R E A A A 0 4 A OR AER 0 6 50 40 30 20 10 0 50 40 30 20 10 0 50 40 30 20 10 0 Time s Time s Time s Cutting Force N Spindle Speed rpm Feed Rate mm min Select Tool Select Operation Basic Advanced v v Reset Quit Figure 16 Final GUI Design The final iteration of the GUI was a significant improvement to all others in terms of functionality multi level usability and style It retained all of the required features from previous iterations such as the six buttons and three real time plots but also added several new features These features include a drop down box for selecting cutting operations button tabs for both basic and advanced users and the ability to create separate controllers with specific parameters linked to a desired pair of operation and tool A full description of this GUI s functionality can be found in section 4 2 1 Results of GUI 3 5 Hardware System Design The system hardware architecture was primarily decided by HNC before the project began although slight modifications had to be made for various reasons Figure 15 below shows the initial hardware provided by HNC Figure 17 Inner Components of the New Contro
26. the best possible outcome Figure 11 below shows the first GUI created FE NESE TE NESE ver Spindle Speed X Force i Feed Rae LJ Ogon oe O Tool Wear Coefficient Y Force Z Force ATTENTION Estimated Hun Tool Wear AO SEE Aa a aio ao Coefficient ieee ROE Sn Ma ee EN eT IEE E Me a Figure 13 Initial GUI Design This GUI met all of HNC s desired functionalities at the time It had start stop reset and alarm off buttons a small alarm screen that flashes in response to severe tool wear Real Time plots of three axis forces are also displayed along with a real time plot of estimated tool wear coefficient These four plots were placed vertically atop one another so that only one time axis was displayed for screen space efficiency Not long after this initial GUI was created HNC greatly changed their requirements shifting the focus of the project to actively controlling feedrate and spindle speed to extend tool life rather than predicting tool wear As a response to these changes a second GUI was made It is shown in Figure 12 below FeedRate xFeedRate yFeedRate spindleSpeed depthOfCut Figure 14 Second Iteration of GUI Design Although iteration 2 of the GUI met the new requirements of displaying the workpiece load and feedrate on real time plots as well as showing actual numerical values of the feedrate and spindle speed it did not display a plot of spindle speed and had limited
27. top tipArea mF gt midRight mF gt midLeft 1l truth 5 ceturn area of trapezoid return totalArea tipArea DONT CALL WITHOUT ABOVE FUNCTION FIRST float getCentroid memberFunction mF temporary floats flo at a b cs get three lengths representative of the trapezoid created in prev function a mF gt midRight mF gt midLeft b mF gt right mF gt left c mF gt midLeft mF gt left Find the x axis centroid of the trapezoid using geometry float result mF gt midLeft 2 a c a a c b a b b b 3 atb return the centroid return result Member Function Diagram class mFD brief Member Function diagram holds 5 members and is used as part of a fuzzyController author Brenden Gibbons Contact btgibbons wpi edu ag allows the use of this C file in CPP classes ifdef cplusplus extern C endif include guards ifndef MFD_H define MFD H Struct to define a MFD membership function diagram 5 entries for all 3 include MemberFunction h typedef struct MFD memberFunction bigNegative The memberFunction associated with big negative values memberFunction negative The memberFunction associated with negative values memberFunction zero The memberFunction associated with middle values memberFunction positive The memberFunction associated with positive values memberFunction bigPositive The memberFunction associated with big
28. 0 Empty to start Now allocate space for the queue entries void arr void calloc max cells sizeof void new queue gt beg arr que beg now points to new queue start new queue gt end amp arr max cells l end now points to end Make the Queue end pointer if new queue gt beg NULL free new queue Unable to allocate queue entries so free struct return NULL new queue gt head new queue gt beg Start at beginning new queue gt tail new queue gt beg start at beginning return new queue int enqueue Queue which queue void ptr Check if queue is already full if which queue gt cells used gt which queue gt max cells which queue gt cells used which queue gt max cells Fixed return 1 Queue overflow check if tail is running off the end if which_queue gt tail gt which_queue gt end wrap around if it is which queue gt tail which queue gt beg enqueue which queue gt tail ptr Store the pointer on the queue which_queue gt tail Point to next free cell which queue gt cells used increment cellsused return 0 Success int forceEnqueue Queue which queue void ptr check if the queue is full if isFull which _queue full so dequeue free float dequeue which queue then enqueue return enqueue which queue ptr else queue not full so simply enqueue return enque
29. 2 4 FBP new_fuzzyController gt feedTable 3 4 FP new fuzzyController gt feedTable 4 4 FP for spindle speed new fuzzyController gt spindleTable 0 0 SN new_fuzzyController gt spindleTable 0 SN new fuzzyController gt spindleTable 2 0 SN new fuzzyController gt spindleTable 3 0 SN new fuzzyController gt spindleTable 4 0 SN new fuzzyController gt spindleTable 0 1 Z new fuzzyController gt spindleTable 1 SN new fuzzyController gt spindleTable 2 1 SN new fuzzyController gt spindleTable 3 1 SN new fuzzyController gt spindleTable 4 1 SN new fuzzyController gt spindleTable 0 2 SP new fuzzyController gt spindleTable 2 SP new fuzzyController gt spindleTable 2 2 Z new fuzzyController gt spindleTable 3 2 SN new fuzzyController gt spindleTable 4 2 SN new fuzzyController gt spindleTable 0 3 SP new fuzzyController gt spindleTable 3 SP new fuzzyController gt spindleTable 2 3 SP new fuzzyController gt spindleTable 3 3 Z new fuzzyController gt spindleTable 4 3 Z new fuzzyController gt spindleTable 0 4 SP new fuzzyController gt spindleTable 4 SP new fuzzyController gt spindleTable 2 4 SP new fuzzyController gt spindleTable 3 4 SP new fuzzyController gt spindleTable 4 4 SP set the other variables new_fuzzyController gt currentSpindleSpeed 1 new _fuzzyController gt currentFeedRate 1 new_fuzzyController gt prevError 0
30. 2 3 1 TMAC by Caron Engineering Founded in 1986 Caron Engineering Inc offers manufacturing and engineering companies a single source for the custom development design and assembly of tool monitoring and adaptive control systems Their product the Tool Monitoring Adaptive Control TMAC system monitors machining horsepower in order to extend cutting tool life It uses slope monitoring on the horsepower to regulate the machines feedrate thus maintaining a constant spindle motor horsepower The TMACs features include a response time of less than 10ms several different machine interfaces preset tool limits for extreme wear and a GUI with real time graphing with the ability to start and stop monitoring 2 3 2 ACM by OMATIVE Systems OMATIVE Systems is a developer of advanced manufacturing and productivity technologies focusing on adaptive control monitoring and metal cutting optimization products They are recognized as an industry and global leader in these technology areas The Adaptive Control Monitor ACM monitors cutting conditions in real time and automatically adjusts the feedrate to its optimal level It uses either maximum load or load band monitoring to optimize tool life and is capable of alarming and stopping the machine in case of tool breakage 2 4 The Computer Numerical Control Milling Process The milling process is a procedure by which a workpiece is fed past a rotating cutting tool in order to cut the workpiece In th
31. 2 for default Fuzzy The PID params float loat loat loat Loat loat Loat Loat fi used fo float cur Loat loat loat loat loat fee a for fuz errorSum loat prevError desiredCu feedKp feedKi feedKd spindleKp spindleKi spindleKd r all rFeedRate currSpindleSpeed controllerGain controllerSignal dRateGain previousCommandSignal zy Error sum for integral term Previous Error to be used for derivative term ttingForce previously determined optimal cutting force Proportional Gain for feed PID Integral Gain for feed PID Derivative Gain for feed PID Proportional Gain for spindle PID Integral Gain for spindle PID Derivative Gain for spindle PID 7 7 7 only used for asynch loat errorBoundary lerror boundary used in asynchronous control mode The current feed rate of the NC The current spindle speed of the NC The controller gain used in calculation The controller signal Gain used when adjusting the feedrate The previous signal sent by the controller Pda fuzzyController theFuzz f f Controlle ah param param param param param param param param E E NE ERE EE return loat desiredSpindleSpeed loat desiredFeedRate float desiredCuttingForce r cGain f cSignal fGain pSignal cRate dSpeed dForce t tool Contro f
32. Diagram ia 13 Figure 7 Sample FUZZY Set tiee NE 14 Figure 8 Fuzzy Logic Spindle Speed Rules Figure 9 Fuzzy Logic Feedrate Rules 16 Figure 10 Input Ne 18 Figure 11 Fuzzy Control Testing LAO 18 Figure 12 Simplified UML DAM antes 21 F ret 13 2 Intl GUL DESI i a 22 Figure 14 Second Iteration of GUIDES iaa 23 Figure 15 Third Iteration of GUI Det 23 Figure 16 Final GUI DO as 24 Figure 17 Inner Components of the New Control System ccc ceeeceeseceneeeeeeeeeeeeeeeeseenneeneeens 25 Figure 18 Hardware System D Madera 26 Figure 19 Synchronous PID Control Response ssranvrenevvrernrnrnvrnvvenvernserrarrrneneesneenseensseresnresnee 30 Figure 20 Asynchronous PID Control Response ccssssscsssceseceseeseeeeeseeseeaceneeeceeeseeseeeeees 31 Figure 21 Spindle Speed Output Fuzzy Seti seasccssesssissevcsisnsassanssseectatasncieassseeiaansadsasssanetanssanre 32 Figure 22 Feedrat Output FU A is 32 Figure 23 Fuzzy Logic Control Response siyicasssscsssssssansessvadasasvaanssetvsascscevsarrentsaadaasbastvssventannsaets 33 Fig r 24 Final GUI De aid 34 Fig r 25 lt GUL Basic Tab sia eae EEA EEEE EEEO EEEE 35 Fre NTN 36 Figure 27 GUI New Tool Parameters Entry n sssersessesssvessresrensnensennsessnranrsevsenrssnsennessenennannssene 36 Figure 28 2 GUI New Operation Entry is 37 Figure 29 GUI Control Algorithm Sele Crm aaa 37 Figure 30 GUI Fuzzy Controller SPK 38 Figure 31 GUI Asynchro
33. In memset amp AIVoltageln 0 sizeof PT AIVoltageln ko Step 3 Set Single end or Differential PR buffer 0x0000 0 single end set device properties f ret DRV DeviceSetProperty fd CFG AiChanConfig if ret DRV_GetErrorMessage ret err msg printf err msg s n err msg DRV_DeviceClose amp fd return 1 Step 2 Config AI Setting Le AIConfig DasChan channel AIConfig DasGain gain ret DRV AIConfig fd amp AIConfig if ret DRV GetErrorMessage ret err msg printf err msg s n err msg DRV_DeviceClose amp fd return 1 if m 0 Binary In AlBinaryIn chan channel AIBinaryIn TrigMode 0 AlBinaryIn reading amp wdata ret DRV AIBinarylIn fd amp AIBinaryIn if ret DRV_GetErrorMessage ret err msg printf err msg s n err msg DRV_DeviceClose amp fd return 1 amp buffer sizeof unsigned int AIScale reading wdata AIScale MaxVolt 5 0 AIScale MaxCount 4095 AlScale offset 2048 AIScale voltage amp voltage ret DRV AIScale fd amp AIScale if ret DRV GetErrorMessage ret err msg printf err msg s n err msg DRV_DeviceClose amp fd return 1 return voltage else if m 1 voltage in AIVoltageIn chan channelNum AlVoltageln gain gain AIVoltageIn TrigMode 0 AIVoltageIn voltage amp voltage ret DRV_AlVoltageln fd amp AIVoltage if ret
34. P err sKp sI errSum sKi sD err prevErr 1 sKd deltaSpindle SP sI sD sum the terms PID i deltaSpindle oe Difference between synchronous and asychrous control With the conditional statement asychronous oe Without the conditional statement synchronous if err gt threshold err lt threshold N N deltaSpindle Update Spindle Speed else V V deltaFeed Update Feedrate end feedVal i V spinVal i N calculate new force f V p N curFor K a f u set error to previous error prevErr err end Graph the results x 1 50 figure 1 plot x feedVal title Feed Rate figure 2 plot x spinVal title Spindle Speed figure 3 p t x y lot x forceVal itle Force label Time 1s label Force N Appendix C Program Code Controller class C brief Contains struct for a controller feedrate ontroller Control functions for both the Spindle speed and author Brenden Gibbons Contact Kl btgibbons wpi edu allows use of this C file in CPP classes tifdef _ cp extern C endif include g ifndef CON define CON include t include lusplus uards TROLLER H TROLLER H ool h FuzzyController h Struct to define a Controller El typedef struct controller unsigned short controllerType 0 for synchPID 1 for asynchPID
35. Queue you want to dequeue from return void head entry of queue NULL if queue is empty Fy void dequeue Queue which queue brief returns a pointer to what the head of the Queue was pointing to param which_queue Queue Pointer to Queue you want to peek return void head entry of queue NULL if queue is empty E void peek Queue which queue brief returns a pointer to what the tail of the Queue was pointing to param which_queue Queue Pointer to Queue you want to peek return void tail entry of queue NULL if queue is empty y void peekTail Queue which queue brief returns an int indicating whether or not the Queue is full param which queue Queue Pointer to Queue return int 1 if full 0 otherwise x7 int isFull Queue which queue Prints an entire queue to standard output param which queue Queue Pointer to queue to print return void el void printQueue Queue which queue tendif ifdef __ cplusplus endif include lt stdlib h gt include lt stdio h gt include home btgibbons MQP queue h Queue create queue int max cells Queue new queue Holds pointer to the newly allocated Queue structure new queue Queue malloc sizeof Queue if new queue NULL return NULL Error unable to allocate Fill in the struct new queue gt max cells max cells new queue gt cells used
36. Signal 0 BRR IKK IR IK RR IR I KKK I I A KKK get the current tool index from the tool index QComboBox theController gt setTool bell clears the dataCollector s queues clears the form s queues sets control parameters to 0 clears curves void Forml reset reset the force feed and spindle arraysex for int i 0 i lt dataSize i t i 0 1 double i time axis yli 0 dummy array force i 0 feed i 0 spindle i 0 if there is a controller reset the prev error and sum error clearOldData theController reset the controller parameters gt desiredSpindleSpeed 0 theController gt currFeedRate 0 theController gt desiredCuttingForce 0 the_fuzzyController gt idealSpindleSpeed 0 the fuzzyController gt idealFeedRate 0 the_fuzzyController gt idealCuttingForce 0 the_fuzzyController gt currentFeedRate 1 the_fuzzyController gt currentSpindleSpeed 1 clear the queues for int i 0 i lt dataSize i forceQueue feedQueue spindleQueue theDataCollector gt xForceQueue theDataCollector gt yForceQueue theDataCollector gt zForceQueue free free free free free free dequeue dequeue dequeue dequeue dequeue dequeue i return stops the timer events void Forml stop printf Stopping Timer n killTimer tehtimer initialize the timer void Forml start
37. Spec Capable Provided CPU AMD Geode 500MHz AMD Geode 500MHz RAM 1GB DDR333SRAM 512MB DDR333SRAM BIOS 4mbit flash bios 4mbit flash bios Compact Up to 128GB Dependent on 1GB Flash OS The specifications of the provided board presented an issue with what operating systems would be installable With only 1 GB of compact flash a typical install of red hat Linux 9 would be impossible so custom options would have to be chosen With a slightly larger 2GB compact flash card this would not be a problem Windows CE has a very small memory footprint 32MB so it could easily be installed on the system however the third factor involved in operating system choice made it an unviable option The final and most influential factor that led to the choice of Red Hat 9 for the operating system was the suggestions of Huazhong Numerical Control The sponsor company suggested using Red Hat 9 to meet the real time requirement of the project The team s advisor Dr Hongqi Liu and his graduate student Zhigang Yang both agreed with this suggestion as they had worked with it previously As a group of non computer scientists the team thought it best to listen to this advice 3 7 Data Collection Methodology The Advantech advDAQ libraries were used for all Data collection The main challenge was to obtain a force value in Newtons from the charge amplifier which outputted voltages To implement this functionality the following steps
38. Truths i FC gt spindleTable j il set the FC currentSpindleSpeed to the centroid of the entire weighted area FC gt currentSpindleSpeed SACSum SASum set the FC currentFeedRate to the centroid of the entire weighted area FC gt currentFeedRate FACSum FASum inline float fuzzAnd float a float b fuzzy AND logic returns the smaller of the two numbers return a gt b b a Member Function class MemberFunction brief Member function used in fuzzy controller Represents a triangle author Brenden Gibbons Contact btgibbons wpi edu xh allows the use of this C file in CPP classes ifdef cplusplus extern C endif include guards ifndef MEMBERFUNCTION H define MEMBERFUNCTION H Struct to define a MemberFunction Only triangle member functions allowed Kf typedef struct MemberFunction float left The x val of left corner of the triangle member float right The x val of right corner of the triangle member float center The x val of the third vertex of the triangle float midLeft Point used in both area and centroid calcs float midRight Point used in both area and centroid calcs memberFunction brief Create and initialize a MemberFunction param left float The left bottom x coord of the member function param right float The right bottom x coord of the member function param mode float The center top x coord of the member func
39. ab file yFqueue outputStringLine file zFqueue for i 0 i lt queuel gt cells used i print the time like a boss outputFloatTab file dequeue from the lqueue loat float malloc sizeof float float dequeue queuel tempf tempf loat outputFloatTab file tempfloat enqueue queuel void tempfloat dequeue from the 2queue print the value tempf tempf loat f loat f loat malloc sizeof float loat dequeue queue2 outputFloatTab file tempfloat enqueue queue2 void tempfloat dequeue from the 3queue print the value tempf tempf loat 1 loat 1 loat malloc sizeof float loat dequeue queue3 outputFloatTab file tempfloat enqueue queue3 void tempfloat dequeue from the 4queue print the value tempf tempf loat f1 loat f loat malloc sizeof float loat dequeue queue4 outputFloatTab file tempfloat enqueue queue4 void tempfloat dequeue from the 5queue print the value tempf loat f1 tempf loat malloc sizeof float loat f1 loat dequeue queued outputFloatTab file tempfloat enqueue queue5 void tempfloat print the value to a file Queue queuel Queue queuel Queue queue6 to to to to a a a a float double i timeStep 1 print the value to a file file file file file then t
40. ach are similar there are many more steps involved in getting an output value for a fuzzy logic controller DesiredState FeedRate a oe Feedback Signal Figure 6 Fuzzy Logic Control Loop Diagram A fuzzy controller is made up of some number of fuzzy sets membership functions to make those sets a set of several rules and truth values on a scale of zero to one An example of a fuzzy set can be seen below in Figure 7 First the error is calculated and is input into a fuzzy set This results in a membership function and a truth value based on where the input lies in the membership function For example a common membership function is in the shape of an isosceles triangle Let s say a particular membership function ranges from 0 to 100 These means if the input is 50 the truth value will be one If the input is 25 or 75 the truth value would be 5 If the input does not fall within this range it will be applied to a different membership function Once this is done a separate input is applied to another fuzzy set in the same manner The membership functions currently in use by each are then applied to the rule set to determine the output A basic rule outline is if A set is B function and C set is D function then output set is E function Determining the truth value to use for the output function depends on how the rule is written There are two main types of rules AND rules and OR rules An AND rule used in the out
41. and z axes and outputs Component proportional electrical charge across three channels Dynamometer Kistler 5070A Charge Takes in the three channel electrical charge output from the Amplifier Dynamometer and converts them to corresponding analog voltages This final hardware system configuration would allow precise control of the HNC210B feedrate and spindle speed sufficient user input capabilities and a large display for the GUI 3 6 Operating System Selection The choice of what operating system to run was mainly determined by three important factors the first of which is as follows The PCM3355 board needed to be able to communicate with the PCM3718HO board through PCM104 pins in order to receive three axis real time force information Advantech only offers the drivers with this functionality for Windows Compact Edition and Red Hat Linux 9 Therefore it was decided that in the interest of time and to prevent complications it was best to use one of these two operating systems as the drivers for data collection were already written Therefore the choice was immediately limited to two operating systems The next factor that played an important role in operating system selection was the limited power and storage of the PCM 3355 board Table 4 below shows the specifications that the board is capable of compared to shows the board s configuration as provided by HNC Table 4 PCM 3355 Specifications
42. ative input of an error block On the positive input of the error block is a constant corresponding to the desired force This block as its name implies calculates the error between the desired and current force This signal along with the derivative signal is sent to a Mux block to combine the data into one signal so it can be interpreted by fuzzy controller This model makes use of two separate fuzzy controllers one for spindle speed and one for feedrate instead of one as described above in the fuzzy methodology section section 3 1 3 This was due to limitations in Simulink that did not allow for multiple output sets in a single controller The two fuzzy logic controls run and calculate their output These outputs are then combined over the next five blocks which calculates the above cutting force equation The simulation continues to loop around until the set run time has completed It is also important to note that with this setup the controller does not begin actively controlling until the second loop 3 3 Software Methodology Before any actual code could be written a requirements analysis for the software was completed This entailed listing the Customer Architectural Structural Behavioral Functional Non Functional and Design requirements All data was obtained in an interview with Dr Hongqi Liu Table 2 below shows this analysis Although most of the following requirements were by the company some additional ones were added to du
43. buttons for user input Its color scheme and layout style also did not match that of HNC s CNCs which was desired Additionally no Quit button was created by accident Iteration 2 was therefore scrapped in favor of the next iteration shown in Figure 13 below bd Formi Cutting Force i 4 F i H 1 4e 05 Al i 1 2e 05 i 1e 05 Te 8e 04 E Ji E 6e 04 E 4e 04 2e 04 4 4 4 02 a t 3 psp NST nm AAA 97 50 40 30 20 10 0 50 40 30 20 10 0 50 40 30 20 10 o Time s Time s Time s Cutting Force 1140 267 N Spindle Speed 5000 000 rpm Feed Rate 128227 422 mm min Select Tool Ideal Spindle Speed 5000 rpm Tool 1 po Initial Feed Rate 5000 mm min Max Force 5000 N Start L Save Data Reset Quit j Figure 15 Third Iteration of GUI Design Iteration 3 of the GUI met all requirements laid out by HNC It contained six user input buttons Start Stop Reset Save Params Save Data and Quit as well as the three required real time plots Additionally a drop down box for cutting tool selection was added for ease There are manual number entry boxes for the experimentally determined maximum allowable cutting force ideal spindle speed and initial feedrate of the cutting operation Although all strict requirements were met one more GUI iteration was created to ensure multi level usability and ease of operation The final iteration of
44. clude tool h Tool create Tool char name float maxForce unsigned short num make a pointer allocate memory for the Tool Tool new_tool new tool Tool malloc sizeof Tool set the properties new tool gt toolName name new tool gt maxAllowableForce maxForce new tool gt number num return return new tool Qt UI file KK KKK KKK KKK KKK KKK KKK KKK KKK K R KKK KKK KKK KKK KKK KKK KKK KKK KKK KKK ui h extension file included from the uic generated form implementation KKK KKK KKK KKK KKK KKK KKK KKK KKK KKK KKK KR KE length of queues and arrays define dataSize 500 timestep of display and control in seconds define timeStep 1 res neccessary imports include queue h include fileIO h include Controller h include DataCollector h include tool h include NCCommunicator h include MemberFunction h include mFD h include FuzzyController h include cuttingoperation h include ParamTable h include lt qstring h gt include lt qwidget h gt Create a queue by allocating a queue structure include lt qapplication h gt include lt qlayout h gt include lt qwt plot h gt include lt qwt plot canvas h gt include lt qwt scale h gt include lt qwt scldraw h gt include lt qwt math h gt include lt qvalidator h gt include lt time h gt include lt string h gt include lt qframe h gt include lt qpainter h gt include lt qcolor h g
45. d float currentFeedRate Holds the last updated feed rate int feedTable 5 5 table for feed fuzzy rules sets int spindleTable 5 5 table for spindle fuzzy rules sets float prevError The previous force error float idealFeedRate Holds the ideal feed rate used in auto creation float idealCuttingForce Holds the ideal feed rate used in auto creation float maxForce force that triggers the alarm fuzzyController brief Create and initialize a fuzzyController param eI mFD The input error mFD param dEI mFD The input change in error mFD param sO mFD The output spindle mFD paramfO mFD The output feed mFD float idealSpindleSpeed Holds the ideal spindle speed used in auto creation return fuzzyController The pointer to the new fuzzyController NULL if fail 2y fuzzyController createfuzzyController mFD el mFD dEl mFD sO mFD 0 brief Create and initialize a default fuzzyController return fuzzyController The pointer to the new fuzzyController NULL if fail el fuzzyController defaultCreateFuzzyController float eNeg float E float ePos float deNeg float DE float dePos float recFeed float recSpindle brief Applies the INPUT fuzzy logic stage and calls applyFuzzyLogic for OUTPUT stage param FC fuzzyController The fuzzyController Xparam forceIn float The current cutting force return void Sets the currentFeedRate and curr
46. database if a lot of tools were used toolArray 0 currTool create Tool Carbide 50000 1 write the dataCollector s queues to text file void Forml saveData check to make sure the queues arent empty if forceQueue gt cells used gt 0 time_t now char the date 50 the date 0 0 now time NULL if now 1 strftime the date 50 Sc txt localtime amp now FILE file fopen the date w printf OPENED THE FILE n writeAllTheQueuesFormatted file forceQueue feedQueue spindleQueue theDataCollector gt xForceQueue theDataCollector gt yForceQueue theDataCollector gt zForceQueue timeStep set controller parameters void Forml setParans LAX he_fuzzyController gt idealSpindleSpeed spindleEdit gt text toFloat the fuzzyController gt currentSpindleSpeed the fuzzyController gt idealSpindleSpeed he fuzzyController gt idealFeedRate feedEdit gt text toFloat the fuzzyController gt currentFeedRate the fuzzyController gt idealFeedRate he fuzzyController gt idealCuttingForce forceEdit gt text toFloat atNaNe theController gt desiredSpindleSpeed spindleEdit gt text toFloat theController gt currFeedRate feedEdit gt text toFloat theController gt desiredCuttingForce forceEdit gt text toFloat theController gt controllerGain 6 theController gt feedRateGain 5 theController gt previousCommand
47. dit gt clear fuzzyDENegEdit gt clear fuzzyDEZeroEdit gt clear fuzzyDEPosEdit gt clear fuzzyRecFeedEdit gt clear fuzzyRecSpindleEdit gt clear fuzzyFrame gt hide enableBackground void Forml disableBackground textLabel7 gt setEnabled false feedShow gt setEnabled false newtonLabel gt setEnabled false mmMinLabel gt setEnabled false pushButton8 gt setEnabled false spindleShow gt setEnabled false rpmLabel gt setEnabled false textlabel6 gt setEnabled false textlLabel5 gt setEnabled false forceShow gt setEnabled false feedPlot gt setEnabled false forcePlot gt setEnabled false spindlePlot gt setEnabled false toolSelectionLabel gt setEnabled false opselectlabel gt setEnabled false buttonsTabs gt setEnabled false opComboBox gt setEnabled false toolComboBox gt setEnabled false forcePlot gt removeMarkers void Forml enableBackground upon tool change textLabel7 gt setEnabled true feedShow gt setEnabled true newtonLabel gt setEnabled true mmMinLabel gt setEnabled true pushButton8 gt setEnabled true spindleShow gt setEnabled true rpmLabel gt setEnabled true textLabel6 gt setEnabled true textLabel5 gt setEnabled true forceShow gt setEnabled true feedPlot gt setEnabled true forcePlot gt setEnabled true spindlePlot gt setEnabled true toolSelectionLabel gt se
48. dle i 0 set up the force plot and curve forcePlot gt setTitle Cutting Force forcePlot gt setAxisTitle QwtPlot xBottom Time s forcePlot gt setAxisTitle QwtPlot yLeft Newtons long forceCurve forcePlot gt insertCurve Force forcePlot gt setCurvePen forceCurve QPen Qt red forcePlot gt setCurveRawData forceCurve t force dataSize forcePlot gt replot forcePlot gt show set up the spindle plot and curve spindlePlot gt setTitle Spindle Speed spindlePlot gt setAxisTitle QwtPlot xBottom Time s spindlePlot gt setAxisTitle QwtPlot yLeft RPM long spindleCurve spindlePlot gt insertCurve Spindle spindlePlot gt setCurvePen spindleCurve QPen Qt blue spindlePlot gt setCurveRawData spindleCurve t spindle dataSize spindlePlot gt replot spindlePlot gt show set up the feed plot and curve feedPlot gt setTitle Feed Rate feedPlot gt setAxisTitle QwtPlot xBottom Time s feedPlot gt setAxisTitle QwtPlot yLeft mm min long feedCurve feedPlot gt insertCurve Feed feedPlot gt setCurvePen feedCurve QPen Qt green feedPlot gt setCurveRawData feedCurve t feed dataSize feedPlot gt replot feedPlot gt show Tools toolComboBox gt insertItem Tool 2 1 toolComboBox gt insertItem Tool 3 1 toolComboBox gt insertItem Tool 4 1 Init the array of tools This would have to be done with a
49. e mFD gt truths 4 getTruth mFD gt bigPositive input float getAreaOfMember mFD mFD float truth int i depending on the memberfunction stated and the truth value ceturn the area if i 1 return getArea mFD gt bigNegative truth if i 2 return getArea mFD gt negative truth if 1 3 return getArea mFD gt zero truth if i 4 return getArea mFD gt positive truth if i 5 return getArea mFD gt bigPositive truth else return 0 float getCentroidOfMember mFD mFD int i depending on the memberfunction stated return the centroid if i 1 return getCentroid mFD gt bigNegative if i 2 return getCentroid mFD gt negative if i 3 return getCentroid mFD gt zero if i 4 return getCentroid mFD gt positive if i 5 return getCentroid mFD gt bigPositive else return 0 NCCommunicator ifdef cplusplus extern C fendif ifndef NCCOMMUNICATOR_H define NCCOMMUNICATOR_H include lt stdlib h gt Struct to define a Controller Ey struct NCCommunicator int comValue int baudRate y typedef struct NCCommunicator NcCommunicator NcCommunicator create NcCommunicator int initCommunications NcCommunicator nc int setSpindleSpeed NcCommunicator nc float rpm int setFeedRate NcCommunicator nc float mmPerMinute int alarmRoutine NcCommunicator nc endif ifdef _ cplusplus e
50. e that the collected data would not match up perfectly with the time step All collected data for each axis is stored in a corresponding queue of a variable specified length This data structure allows a constant flow of incoming data to be easily processed and displayed versus time Summary In summary three separate control methods were proposed including fuzzy synchronous PID and asynchronous PID Tests were then designed for these algorithms Next a data collection scheme was created and a suitable operating system was selected Finally a software architectural structure was conceived and a comprehensive and user friendly GUI was designed 4 Results The following section describes the final outcome of the project It outlines what was completed and how it corresponds to the initial goals laid out in the Introduction section 4 1 Control Algorithms Although the controller was not able to be tested in a real life scenario all three implemented control algorithms were able to be simulated and perform quite well under the theoretical conditions after some fine tuning In the next sections the performance of each control algorithm along with sample results will be discussed 4 1 2 Proportional Integral Derivative Control System Development Figure 17 below displays the response of the synchronous PID to the sample data discussed is section 3 2 Force 750 700 650 600 Force N 550 500 450 1
51. ecSpindle cont gt desiredSpindleSpeed recSpindle cont gt desiredFeedRate recFeed void multiControl Controller cont float currForce double timestep if cont gt controllerType 0 synchronousControlFunction cont currForce timestep else if cont gt controllerType 1 asynchronousControlFunction cont currForce timestep else if cont gt controllerType 2 Control cont gt theFuzz currForce cont gt currSpindleSpeed getNormalizedSpindleSpeed cont gt theFuzz cont gt currFeedRate getNormalizedFeedRate cont gt theFuzz void asynchronousControlFunction Controller cont float currForce double timestep get the force error float error cont gt desiredCuttingForce currForce temp float calculation variables float pVal iVal dVal sum the error cont gt errorSum error cont gt errorSum check whether to adjust the Spindle speed or Feed Rate if error gt cont gt errorBoundary error lt cont gt errorBoundary 1 large error so adjust the Feed Rate pVal error cont gt feedKp iVal cont gt errorSum cont gt feedKi dVal float error cont gt prevError timestep cont gt feedKd cont gt currFeedRate pVal iVal dVal else small error so adjust the Spindle Speed pVal error cont gt spindleKp iVal cont gt errorSum cont gt spindleKi dVal float error cont gt prevError timestep cont gt spindleKd cont gt currSpind
52. ector gt mode m printf The entered mode is d n m printf The recorded mode is is d n new dataCollector gt mode set the Queues new dataCollector gt xForceQueue create queue size new dataCollector gt yForceQueue create queue size new dataCollector gt zForceQueue create queue size make sure the Queues were allocated if new_dataCollector gt xForceQueue Inew dataCollector gt yForceQueue Inew dataCollector gt zForceQueue printf Empty Queues return NULL new_dataCollector gt lastvalue 1000 return new dataCollector int makerand int high return rand high 1 returns a random number from 0 to high int pullAnalogPort dataCollector the dataCollector int channelNum int m if we are not in computer simulation we must pull from the appropriate port This can t run on the computer so its commented out Step 1 Open Device PT AIConfig AIConfig PT AIBinaryIn AIBinaryIn PT AIVoltageIn AIVoltageln PT AIScale AIScale unsigned short wdata unsigned short channel unsigned short gain unsigned int buffer float voltage 0 char err msg 100 char filename dev advdaq0 int ret int fd ret DRV DeviceOpen filename amp fd if ret DRV GetErrorMessage ret err msg printf err msg s n err msg return 1 memset amp AIConfig 0 sizeof PT AIConfig memset amp AIBinarylIn 0 sizeof PT _AIBinary
53. eedrate the PID controller needs two different sets of gains one set of gains for the feedrate and the other for spindle speed This allows for more accuracy and higher efficiency from the control function This is because the two parameters affect the cutting force in slightly different ways Altering two different variables also allows for the use of two different versions of a PID One version that controls the spindle speed and feedrate simultaneously a synchronous controller and one that controls the two values separately an asynchronous controller The asynchronous controller manipulates spindle speed for larger errors Once the error crosses a predetermined threshold control switches to varying the feedrate for the final changes Both versions were implemented in order to give the user a bit more choice and control for cutting operations depending on the desired response The two versions of the PID can be viewed in MATLAB code in Appendix B It should be noted that this form alters slightly from the version in the actual program as this was also used for testing purposes which will be discussed shortly 3 1 3 Fuzzy Logic Methodology As discussed in the fuzzy logic background section 2 6 2 fuzzy controllers have been implemented to control cutting force through spindle speed and feedrate changes successfully These articles were used as a guide for the implementation of this controller The first step was to determine the overall layou
54. em to all functions modifying them All code would be written in the ANSI C99 standard to ensure portability between compilers For planning purposes a UML Unified Modeling Language diagram was created This allowed separate functions in each class to be specifically defined before coding and also a visualization of how incoming data travels through and is modified in each of the modules Several iterations of the UML diagram were created both before and during coding The iterations were necessary when certain requirements could not easily be met with the UML diagram at the time A simplified version of the latest and most complete iteration of the UML diagram is shown below in Figure 10 The full version can be found in Appendix A NCCommunicator filelO dataCollector gt Main Gui K gt Controller gt FuzzyControtter y 9 queue mFD paramTable koH cuttingOperation Tool memberF unction Figure 12 Simplified UML Diagram Many measures were taken to ensure that the code base was easily expandable The first of these was the heavy compartmentalization of the code Separate C modules header and c file were created for almost every task in code In all nine individual purpose modules were written and a tenth for the GUI The separation of tasks into modules also allowed ease of testing and debugging for each section of code The second measure wa
55. en Feed Rate is fZ 1 14 If Force is ZE and Delta Force is cP then Feed Rate is fP 1 15 If Force is ZE and Delta Force is cBP then Feed Rate is fBP 1 16 If Force is P and Delta Force is cBN then Feed Rate is fBN 1 17 If Force is P and Delta Force is cN then Feed Rate is fN 1 18 If Force is P and Delta Force is cZ then Feed Rate is fN 1 19 If Force is P and Delta Force is cP then Feed Rate is fZ 1 20 If Force is P and Delta Force is cBP then Feed Rate is fP 1 21 If Force is BP and Delta Force is cBN then Feed Rate is fBN 1 22 If Force is BP and Delta Force is cN then Feed Rate is fBN 1 23 If Force is BP and Delta Force is cZ then Feed Rate is fN 1 24 If Force is BP and Delta Force is cP then Feed Rate is fZ 1 25 If Force is BP and Delta Force is cBP then Feed Rate is fP 1 Figure 9 Fuzzy Logic Feedrate Rules 3 2 Testing the Control Algorithms A control algorithm cannot be implemented without first testing it In a real scenario the controller would send the mill some value the mill would apply the change and a new force would result This new force would be sent back into the control algorithm to continue making adjustments Because the test is a simulated scenario there is no live update of the cutting force that happens as a natural progression Because of this a separate function had to be implemented to calculate what the resultant
56. entSpindleSpeed Kol void Control fuzzyController FC float forceln brief Applies the OUTPUT stage of fuzzy control param FC fuzzyController The fuzzyController param errorTruths float array of error truths param deltaErrorTruths float array of change in error truths return void Sets the currentFeedRate and currentSpindleSpeed ay void applyFuzzyLogic fuzzyController FC float errorTruths float deltaErrorTruths brief Return the Feed Rate to be sent to the mill param FC fuzzyController The fuzzyController to get the feed rate of return float The normalized feedrate Kil float getNormalizedFeedRate fuzzyController FC brief Return the Spindle to be sent to the mill param FC fuzzyController The fuzzyController to get the spindle speed of return float The normalized spindle speed L float getNormalizedSpindleSpeed fuzzyController FC Inline function for ANDing fuzzy logic style inline float fuzzAnd float a float b tendif ifdef cplusplus tendif include lt stdlib h gt include lt stdio h gt include FuzzyController h fuzzyController createfuzzyController mFD eI mFD dEI mFD sO mFD 0 pointer to the new mFD fuzzyController new fuzzyController allocate mem for mFD new fuzzyController fuzzyController malloc sizeof fuzzyController if new fuzzyController NULL return NULL Error unable to all
57. ers Neil and Hermione for helping with the project and assisting us with everyday life in China Table of Contents AR 2 Ackiowledgem rts vusesseuStaadhsveasssucunatenadicorsuasuotinesiudscae iii 3 dm 8 2 Background A ais 9 2 1 Company Background EEE EE 9 2 2 Orginal System EE RE EN 9 23 COMPE in 10 2 3 1 TMAC by Caron Engineering ide 10 2 3 2 ACM DY OMA TIVE Syste Sreser iesim eree iii 10 2 4 The Computer Numerical Control Milling Process rerrnrnvnnnvrnvrenrvererrrernverneenseenevereenrennee 10 2 3 Past RESEARCH AA 12 2 6 Control Alcorisa 12 2 6 1 Proportional Integral Derivative Cono laa d 12 2 6 2 Fuzzy Logic fr Luse eskere Ioannes 13 Sy VISTO DONO A iia 15 3 1 Control Algorithms Methodology x scissssnsiatssuvecnsadeacenedaniiexesanndancantheutagendanassnidexwioneaeienions 15 3 1 2 Proportional Integral Derivative Methodology ooooonccnnccnociooccoccnononononcnnnncnna conan nonnnonos 15 3 1 3 Fuzzy Logic Med iaa 15 3 2 Testing the Control Alsos a cd 17 3 2 1 Proportional Integral Derivative Control Testing ceeeeeseceseeneeeeeeeeeeeceseenseeneeees 17 22 F zzy L se Testing se due 18 3 3 Software Methodology nai 19 3 4 Graphical User Interface Design iassssicsssssiccasantncsnnesdnccensanedesnassaeceassssasnaasanerseiasndeissesniveatasocss 21 3 5 Hardware Systemi Eee 25 3 0 GE System SO ii es 21 3 7 Data Collection Methoddl y ussassnesansnmmmssnuvjimmunaissuntnj iavhoad nainn 28 nn o LEO ee eg TEET 29 A A 30 4 1
58. force read by the 3 axis sensor return float returns the updated feed rate to be sent to the NC void asynchronousControlFunction Controller cont float forceError double timestep brief Control function for determing the correct Spindle Speed param cont Controller A pointer to the controller in question param currForce float The current cutting force read by the 3 axis sensor return float returns the updated spindle speed to be sent to the NC S void synchronousControlFunction Controller cont float forceError double timestep called when changing controllers or resetting to clear past sums void clearOldData Controller cont fendif ifdef _ cplusplus fendif include lt stdlib h gt include lt stdio h gt include Controller h Controller create Controller Controller new_controller Holds pointer to the newly allocated Controller structure new controller Controller malloc sizeof Controller if new controller NULL return NULL Error unable to allocate Fill in some of the struct new_controller gt controllerType 3 new controller gt errorSum 0 new_controller gt prevError 0 return new_controller void init_as_asynchPID Controller cont float desF float fkp float fki float fkd float skp float ski float skd float errbound cont gt controllerType 1 asynch cont gt desiredCuttingForce desF cont gt feedK
59. given tool An example would be 3D contouring ANSI 6160 Aluminum or Grooving generator shaft Once a new operation is created it is added to the drop down box of operations on the left hand side of the GUI allowing it to be selected whenever desired just like a tool There is a maximum of 25 unique operations after which a message indicating that no more new ones may be created is shown Enter New Operation Name Operation Name Save Operation Cancel Figure 28 GUI New Operation Entry The New Controller button can only be pressed when both a Tool and an Operation have been selected from their drop down menus When pressed it pops up a window for selecting the type of new controller the user wishes to create As shown below in Figure 27 one of three controller types may be selected Synchronous PID Asynchronous PID or Fuzzy Asynchronous PID Figure 29 GUI Control Algorithm Selection Select Controller Type In each instance the new controller is bound to the currently selected Tool Operation combination This means that it will be saved upon creation and automatically selected in the future whenever that Tool Operation combination is selected from the drop down menus This feature allows advanced users to set up complicated parameters while allowing basic users to operate the program by remaining in the Basic tab The Create new Fuzzy Controller dialog pops up when the Fuz
60. gt toolName 1 cancelNewTool changeTool setEnabled true void Forml saveOperation QString strl operationNameEdit gt text copy QString to char Ghar cstr std string fname fname strl ascii cstr new char fname size 1 strepy cstr fname c str cuttingOperation aNewOperation create cuttingOperation cstr operationIndex operationArray operationIndex aNewOperation operationIndex if too many tools then say it if operationIndex gt 25 maxOperationsReachedLabel gt show newOperationButton gt setEnabled false newToolFrame gt setIsPopup false opComboBox gt insertItem aNewOperation gt operationName 1 cancelNewOperation changeOperation setEnabled true void Forml popSynchPIDFrame controllerTypes gt hide synchPIDFrame gt raise synchPIDFrame gt show synchPIDFrame gt repaint void Forml saveSynchPID make a new controller and pointer to the controller Controller newCont create Controller init the controller as a SynchPID with inputted params init as synchPID newCont newRecForceEdit gt text toFloat newSynchfKpEdit gt text toFloat newSynchfKdEdit gt text toFloat newSynchfKiEdit gt text toFloat r newSynchsKpEdit gt text toFloat newSynchsKdEdit gt text toFloat r r newSynchsKiEdit gt text toFloat Di
61. has been written to implement this feature however it has not been tested due to unsuccessfully setting up the hardware 6 2 Communication with the Numerical Controller Assuming the proper documentation can be found the dummy function currently implemented would have to be replaced by the proper protocols The function calls are already implemented in the correct places so replacing the protocols is all that would need to be done 6 3 Live Testing Assuming Red Hat 9 can be successfully installed with a larger compact flash card the entire system would have to be tested and debugged Some error would likely occur and these would have to be addressed 6 4 Database Implementation An SQL database could be used as a more efficient storage method for individual tool and cutting operation information and sets This would also allow tools that have been created to persist in the event that the program is closed 6 5 Load Band Monitoring In the current implementation the system uses an experimentally determined maximum cutting force value to decide when to trigger the alarm Another option would be to use load band monitoring for deciding when to trigger the alarm This would allow for early warnings in the case of imminent tool breakage This would be easily implementable in the code base Works Cited l Koren Yoram Adaptive Control Systems for Machining Manufacturing Perspective 2 1 1989 Web Autumn 2011 lt http
62. he rule set 2 However because they used seven membership functions per fuzzy set their rule sets didn t directly translate and adjustments had to be made ek 8 and 9 below el the rules used for the controller 2 If Errori is BN and Change_Error is N then Spindle_Speed is Z 1 3 If Error is BN and Change_Error is cZ then Spindle_Speed is sP 1 4 If Error is BN and Change Error is cP then Spindle Speed is sP 1 5 If Error is BN and Change_Error is cBP then Spindle_Speed is sP 1 6 If Error is N and Change Error is cBN then Spindle Speed is sN 1 7 If Error is N and Change Error is cN then Spindle Speed is sN 1 8 If Error is N and Change Error is cZ then Spindle Speed is sP 1 9 If Error is N and Change Error is cP then Spindle Speed is sP 1 10 If Error is N and Change Error is cBP then Spindle Speed is sP 1 11 If Error is Z and Change Error is cBN then Spindle Speed is sN 1 12 If Error is Z and Change Error is cN then Spindle Speed is sN 1 13 If Error is Z and Change Error is cZ then Spindle Speed is Z 1 14 If Error is Z and Change Error is cP then Spindle Speed is sP 1 15 If Error is Z and Change Error is cBP then Spindle Speed is sP 1 16 If Error is P and Change Error is cBN then Spindle Speed is sN 1 17 If Error is P and Change Error is cN then Spindle Speed is sN 1 18 If Error is P and Change Err
63. hen then then then re enqueue re enqueue re enqueue re enqueue re enqueue then re enqueue Queue queue2 double timeStep so only need to for with 1 dequeue from the 6queue print the value to a file then re enqueue tempfloat float malloc sizeof float tempfloat float dequeue queue outputFloatLine file tempfloat enqueue queue6 void tempfloat folose file return 1 success Fuzzy Controller class FuzzyController brief A fully customizable fuzzy controller for use in the GUI author Brenden Gibbons Contact btgibbons wpi edu allows the use of this C file in CPP classes ifdef cplusplus extern C fendif include guards ifndef FUZZYCONTROLLER H define FUZZYCONTROLLER H defines for member functions define EBN 100 define DEBN 100 define SBN 1 define FBN 1 define EN 50 define DEN 50 efine SN 2 efine FN 2 efine Z 3 efine EP 50 efine DEP 50 efine SP 4 define FP 4 define EBP 100 define DEBP 100 define SBP 5 define FBP 5 q q q q q q include mFD h Struct to define a FuzzyController AG typedef struct FuzzyController mFD errorInputMFD The mFD for input error mFD deltaErrorInputMFD The mFD for input delta error mFD spindleOutputMFD The mFD for output spindle mFD feedOutputMFD The mFD for output spindle float currentSpindleSpeed Holds the last updated spindle spee
64. is process the cutting tool is held horizontally stationary while translating vertically to increase or decrease the depth of cut The rate at which the cutting tool is spinning is called spindle speed Increasing the spindle speed can reduce the cutting force and increase the quality of the cut The workpiece is held in place on a horizontally translating table and is fed past the cutting tool The rate at which the workpiece moves across the cutting tool is called the feedrate Feedrate has the opposite effect of spindle speed in that decreasing it will reduce the cutting force and increase the quality of the cut A picture of a mill displaying the spindle and table can be seen below in Figure 3 Figure 3 Milling Machine Table and Spindle In many cases this process is controlled by a CNC or computer numerical controller This is a computer which controls all aspects of the cutting process and allows accuracy and speed beyond that of manual control A picture of the HNC 2101B the project specific CNC can be seen below in Figure 4 HARE Be JARSE eee poo EE Opa po on 562 gan 0 Ea Figure 4 Milling Machine Computer Numerical Controller 2 5 Past Research Monitoring of tool wear and breakage on milling machines often referred to as adaptive control in industry has been a topic of research for quite some time and several different methods of control have been developed over the years There are three main methods of ada
65. its members and initializes it param t float the Controller gain of the new Controller param mat float The controller signal param maxForce float Gain used when adjusting the feedrate param refForce float The previous signal sent by the controller return Controller returns a pointer to the newly allocated Controller XL cuttingOperation create cuttingOperation char name unsigned short num fendif ifdef cplusplus fendif include lt stdlib h gt include lt string h gt include cuttingoperation h cuttingOperation create cuttingOperation char name unsigned short num cuttingOperation new_operation new operation cuttingOperation malloc sizeof cuttingOperation new operation gt operationName name new_operation gt number num return new_operation Data Collector class DataCollector brief Contains struct for a dataCollector and functions for collecting data author Brenden Gibbons Contact btgibbons wpi edu Kf allows use of this C file in CPP classes ifdef _ cplusplus extern C tendif include guards ifndef DATACOLLECTOR_H define DATACOLLECTOR_H include queue h Struct to define a Controller Ey typedef struct DataCollector int samplingRateHz the AD sampling rate Hz int mode 1 for sim 0 for real Queue xForceQueue pointer to the z Force Queue Queue yForceQueue pointer to the
66. kground the combo already existed else TODO pop a window that a Controller has already been made printf A controller already exists n void Forml popNewOperationFrame cancelNewTool opComboBox gt setEnabled false opComboBox gt clearFocus newToolFrame gt setIsPopup true newOperationFrame gt raise newOperationFrame gt show newOperationFrame gt repaint disableBackground void Forml cancelNewTool toolNameEdit gt clear maxForceEdit gt clear newToolFrame gt hide toolComboBox gt insertItem Tool 1 1 enableBackground void Forml cancelNewOperation operationNameEdit gt clear newOperationFrame gt hide enableBackground void Forml saveTool QString strl toolNameEdit gt text copy QString to char Ch r cstx std string fname fname strl ascii cstr new char fname size 1 strepy cstr fname c str Tool aNewTool create Tool cstr maxForceEdit gt text toFloat toolIndex toolArray toolIndex aNewTool add tool to tool array Tool newTool create Tool toolNameEdit gt text maxForceEdit gt text toFloat toolIndex toolArray toolIndex newTool toolIndex if too many tools then say it if toolIndex gt 25 maxToolsReachedLabel gt show newToolButton gt setEnabled false newToolFrame gt setIsPopup false toolComboBox gt insertItem aNewTool
67. l System It consisted of a power supply a PCM3718HO board a PCM 3778 board a 5 TFT LCD display five buttons for user input and a small fan inside of a steel and plastic case Upon initial inspection many wires appeared to be broken or disconnected in potentially harmful ways To avoid serious problems datasheets for all components were analyzed to determine exactly what wires needed to be replaced connected to what A total of three wires needed to be fixed as well as a plastic end cap used to connect analog keyboard input to the PCM3778 Despite limited resources these repairs were made successfully Once a keyboard could successfully be connected installation of the operating system began A problem arose when connecting the PCM3778 board to the TFT LCD display Although all wires were undamaged the display would only turn on about two percent of the time and only for moments This problem was traced to a faulty PCM 3778 board so it was replaced with a newer model with similar capabilities the PCM 3355 Once installed the new board also did not work with the TFT LCD display For diagnostic purposes a normal LCD desktop monitor was hooked up and tested It worked successfully indicating that the TFT LCD itself was malfunctioning Although the Desktop LCD Display could interface with the new PCM 3355 board the board did not have any firmware or BIOS installed by the factory rendering it unusable until the proper manufacturer s CD
68. leSpeed pVal iVal dVal cont gt prevError error void synchronousControlFunction Controller cont float currForce double timestep get the force error float error cont gt desiredCuttingForce currForce printf desF f n cont gt desiredCuttingForce printf currF f n currForce printf Error f n error temp float calculation variables float pVal iVal dVal sum the error cont gt errorSum error cont gt errorSum Adjust both the Feed rate and the Spindle speed adjust Feed Rate pVal error cont gt feedKp iVal cont gt errorSum cont gt feedKi dVal float error cont gt prevError timestep cont gt feedKd cont gt currFeedRate pVal iVal dVal adjust the Spindle Speed pVal error cont gt spindleKp iVal cont gt errorSum cont gt spindleKi dVal float error cont gt prevError timestep cont gt spindleKd cont gt currSpindleSpeed pVal iVal dVal cont gt prevError error void clearOldData Controller cont cont gt prevError 0 cont gt errorSum 0 Cutting Operation ifdef cplusplus extern C fendif ifndef CUTTINGOPERATION H define CUTTINGOPERATION H Struct to define a tool and related functions Ey typedef struct cuttingoperation char operationName unsigned short number for easy ID cuttingOperation brief Creates a tool by allocating mem for the struct and
69. line above compares the two different truth values and uses the lower of the two An OR rule will use the higher of the two values It is likely that given just two inputs there will be many more than just one output as membership functions generally overlap resulting in multiple rules being met These outputs are all applied to the same output set Once all rules have been checked and applied the center of gravity or centroid is found based on weighing all of the individual outputs The position of the centroid along the x axis is the final output of the system T T T T er BN N ZE P BP Figure 7 Sample Fuzzy Set 3 Methodology The following section describes how the project was approached specifically what techniques and practices were used It shows how the project was taken from a concept to an actual product 3 1 Control Algorithms Methodology For the adaptive controller two control algorithms were chosen to be implemented a PID proportional derivative integral controller and a fuzzy logic controller The PID was chosen as it is a proven and easily implementable method The fuzzy logic controller was chosen as it provides rapid error correction and tends to be a bit more forgiving than a PID controller Each algorithm will be discussed further in their respective sections below 3 1 2 Proportional Integral Derivative Methodology Because the controller has to alter two different values spindle speed and f
70. ller Create New Synchronous PID Controller Recommended Force Feed Kp Feed Kd Feed Ki Spindle Kp Spindle Kd Spindle Ki Figure 32 GUI Synchronous PID Controller Setup If the cutting force maximum value specified for each tool upon creation is ever exceeded a stop command is issued to the Numerical Controller within 10ms Additionally an Alarm screen seen below in Figure 31 is displayed which encompasses the entirety of the GUI Data Collection and plotting are immediately halted and adaptive control of the feedrate and spindle speed is suspended All current stored data is saved to file to permit later inspection of the error The Alarm screen strobes between bright red and black and displays a highly contrasting warning message The Reset button on the lower right hand side allows the user to end the alarm Figure 33 GUI Alarm Screen 4 4 Data Collection Results All of the Advantech DAQ boards use the same API for collecting data ActiveDaq This means that if a program is able to successfully acquire data from one Advantech DAQ board it will also succeed on any other board As the main program was never successfully run on the PCM3355 board the program to collect data from the Kistler 5070A charge amplifier was tested on an Ubuntu computer using an Advantech USB 4718 Data was successfully collected in this manner The results are shown in Figure 32 below 200 root il
71. named with a timestamp for ease of filing The first column displays the time increment The second third and fourth columns show the cutting force spindle speed and feedrate respectively The final three columns are the x y and z forces as measured by the force plate The file is stored in the same directory as the program is run from This functionality was one of the specific details communicated by HNC The New Tool Dialog shown below in Figure 25 is evoked when the New Tool button is pressed This dialog allows users to create tools with specified names and maximum cutting forces The maximum cutting force value is an experimentally determined value for each tool over which a tool break will likely occur The Alarm is triggered when cutting force exceeds this value Once a new tool is created it is added to the drop down box of tools on the left hand side of the GUI allowing it to be selected whenever desired There is a maximum of 25 unique tools after which a message indicating that no more new ones may be created is shown Enter New Tool Parameters Tool Name Max Force Save New Tool Cancel Figure 27 GUI New Tool Parameters Entry New Operation The New Operation Dialog Figure 26 below is conjured when the button is pressed This dialog allows users to create operations with specified names Each new operation created represents an action one would perform using a CNC machine using a
72. nction memberFunction new_memberFunction allocate mem for datacollector new_memberFunction memberFunction malloc sizeof memberFunction if new memberFunction NULL return NULL Error unable to allocate set all the members new_memberFunction gt left left new_memberFunction gt right right new_memberFunction gt center center these start as 0 because we can know them without a truth value new_memberFunction gt midLeft 0 new_memberFunction gt midRight 0 return new_memberFunction for input phase float getTruth memberFunction mF float input if the input is within the memberfunction range if input gt mF gt left amp amp mF gt right gt input return the associated truth return 1 fabs mF gt left mF gt right 2 input mF gt right mF gt left else the truth is 0 else return 0 for output phase float getArea memberFunction mF float truth make temporary floats float totalArea tipArea slopeLeft slopeRight first get the area of the whole triangle totalArea 5 mF gt right mF gt left find the slopes of the two edges slopeLeft 1 0 mF gt center mF gt left slopeRight 1 0 mF gt right mF gt center find intersection of both edges with line y truth mF gt midLeft mF gt left 1 0 slopeLeft truth mF gt midRight mF gt right 1 0 slopeRight truth get the area of the small triangle at the
73. nd to the forces 5 6 As the PCM 3718HO DAQ board s analog input ports can only measure changes in voltage a Kistler 5070A charge amplifier was used to convert the dynamometer s output of electrical charge to a corresponding change in voltage 4 5 CNC Communication Results Although information about communicating on the fly with other companies numerical controllers through RS32 or Ethernet could be found no such material from HNC was available Due to time constraints and the aforementioned this task of the project was left uncompleted 4 6 Summary Overall the three control algorithms conceptualized in the methodology section were implemented and tested in MATLAB Data was successfully collected from and verified with an oscilloscope Finally an appropriate software system was developed in C and C to meet most of the project goals and a user friendly GUI was created 5 Conclusion Overall the project was a success Four out of six of the objectives laid out in the introduction were met and another was partially met A viable replacement for HNC s current machining tool wear and breakage monitor that operates in soft real time and implements a tool life extending spindle speed and feedrate control system was produced This system is able to detect significant machining tool abrasion and breakage through the analysis of cutting forces extend tool life by actively controlling feedrate and spindle speed continuously dis
74. ndif include lt stdlib h gt include lt stdio h gt include NCCommunicator h include Controller h NcCommunicator create NcCommunicator NcCommunicator tehpointer Holds pointer to the newly allocated Queue structure tehpointer NcCommunicator malloc sizeof NcCommunicator return tehpointer int initCommunications NcCommunicator nc printf Initializing Communications return 1 int setSpindleSpeed NcCommunicator nc float rpm printf Setting Spindle Speed to f rpm rpm return 1 int setFeedRate NcCommunicator nc float mmPerMinute printf Setting Feed Rate to f mm min mmPerMinute return 1 7 int alarmRoutine NcCommunicator nc set spindle speed to zero setSpindleSpeed nc 0 return 1 Cutting Operation include lt stdlib h gt include lt string h gt include cuttingOperation h cuttingOperation create _cuttingOperation char name unsigned short num cuttingOperation cuttingOperation new operation cuttingOperation malloc sizeof cuttingOperation new_operation gt operationName name new operation gt number num return new_operation ParamTable class ParamTable author Brenden Gibbons Contact btgibbons wpi edu EN allows use of this C file in CPP classes ifdef cplusplus extern C fendif include guards ifndef PARAMTABLE H define PARAMTABLE H include tool h include cuttingopera
75. nnovative and reliable systems to its customers HNC s main products are their CNC controllers however they also offer servo drivers and motors CNC machines and infrared products HNC products are used worldwide in a variety of different businesses such as the textile mechanical chemical medical and steel industries HNC also uses their products for educational services such as technical personnel training 2 2 Original System The currently implemented system which can be seen below in Figure 2 is a black box with a screen on the top half and buttons to control the program on the bottom half In the bottom half of the picture is a screenshot of the system while on but not currently monitoring anything As can be seen there is only one area to display information about the cutting force and a list of options on the right side This setup is not very user friendly and also is not capable of displaying all of the desired information Figure 2 Top Original Control System Bottom User Interface On top of a poor user interface the software of the system is outdated The system runs on DOS which does not allow for real time operation Also the signal processing algorithms being used are slow and do not allow for optimal performance Unfortunately there was no specific information was available about the controller to allow for a more thorough analysis 2 3 Competitors In this section two competitive products are discussed
76. nous PID Controller Setup 0 0 0 e ce eeceseeceeseceseceeeeneeeeeeeeeeenseenaeeneeens 39 Figure 32 GUI Synchronous PID Controller Setup ainia ce 39 Figure 33 GUI Alarm SEE 40 Figure 34 Data Collection Testing Results ai 41 Figure 35 Data Collection Testing Input Square Wave ssrssssesnseennsennvsannnsansssnnsensvssnnssenssnnnsses 41 Table of Tables Table 1 Cutting Force Equation Historical Trends ssrernvnrrnrronrrrnnrnrnrrrerrrensrrrnrerrrnrersrennessnneen 17 Table 2 Reguitements ANS AA E SEAE EREA 19 Table 3 Hardware Components ia 26 Table 4 PCM 3355 Pe SE 27 Table 5 Synchronous PID Gain Valls ii 30 Table 6 Asynchronous PID Control Gain Values sia is 31 1 Introduction The CNC milling process involves removing material from a workpiece by feeding it past a rotating cutting tool The process inevitably leads to tool abrasion and therefore increased cutting forces which will eventually cause tool breakage This tool breakage can result in a ruined workpiece as well as decreases in efficiency As such an effective tool wear monitor and CNC controller is extremely useful in milling Figure 1 below displays the rotating shaft of a large generator of which the center portion is cut using a mill Milling this piece was the original scope of this project as the ramifications of a broken tool and therefore a broken shaft are much greater than in most other cases Figure 1 Large Generator Shaft A tool wear
77. now save the controller to the Ptable addSet the Ptable newCont toolComboBox gt currentItem opComboBox gt currentItem closeSynchPID clearOldData theController theController newCont enable the start button startButton gt setEnabled true void Forml closeSynchPID 7 newSynchfKpEdit gt clear newSynchfKdEdit gt clear newSynchfKiEdit gt clear newSynchsKpEdit gt clear 7 7 newSynchsKdEdit gt clear newSynchsKiEdit gt clear newRecForceEdit gt clear synchPIDFrame gt hide enableBackground 7 7 void Forml popAsynchPIDFrame controllerTypes gt hide asynchPIDFrame gt raise asynchPIDFrame gt show asynchPIDFrame gt repaint void Forml saveAsynchPID make a new controller and pointer to the controller Controller newCont create Controller init the controller as an AsynchPID with inputted params init_as_asynchPID newCont newAsynchRecForceEdit gt text toFloat newAsynchfKpEdit gt text toFloat newAsynchfKdEdit gt text toFloat newAsynchfKiEdit gt text toFloat newAsynchsKpEdit gt text toFloat newAsynchsKdEdit gt text toFloat newAsynchsKiEdit gt text toFloat AsynchCutoffForce gt text toFloat Di now save the controller to the Ptable addSet the Ptable newCont toolComboBox gt currentItem opComboBox gt currentItem clearOldData theController theController ne
78. oat font gothroa Member nFD float ing font getlentnoidd tember mF D int NeConmunicator create Nc Communicator tinind Communications void tntsaFoodRatoioat Appendix B Matlab Test Code u 87 SUser Input a 4 User Input depth of cut mm p 4 User Input number of teeth K 500 User Input specific cutting force N 2000 Starting Spindle Speed V 2262 sStarting feedrate threshold 15 User Input cutting force threshold used for asynch refFor 500 User Input desired force SStarting cutting forces xForce 600 yForce 400 zForce 100 err 0 current error prevErr 0 previous error errSum 0 sum of errors Feedrate Gains fKp 1 SUser Input fKi 01 User Input fKd 05 User Input oe Spindle Speed Gains sKp 1 3 SUser Input sKi 03 User Input sKd 01 User Input Testing arrarys feedVal zeros 1 50 spinVal zeros 1 50 forceVal zeros 1 50 PID zeros 1 50 error zeros 1 50 SMagnitude of cutting force curFor sqrt xForce 2 yForce 2 zForce 2 Sloop 50 times for i 1 50 forceVal i curFor err refFor curFor calculcate error error i err errSum err errSum calculate sum of errors Calculate feedrate PID terms fP err fKp fI errSum fKi D err prevErr 1 fKd deltaFeed fP fI fD sum the terms SCalculate spindle speed PID terms s
79. ocate set the mFDs new fuzzyController gt errorInputMFD el new fuzzyController gt deltaErrorInputMFD dEIL new fuzzyController gt spindleOutputMFD s0 new fuzzyController gt feedOutputMFD f0 here is where all the fuzzy sets are defined in fuzzy tables These are based on values from Fuzzy Control of spindle power in end milling process for Feeds new fuzzyController gt feedTable 0 0 FN new fuzzyController gt feedTable l 0 FN new fuzzyController gt feedTable 2 0 FBN new fuzzyController gt feedTable 3 0 FBN new fuzzyController gt feedTable 4 0 FBN new fuzzyController gt feedTable 0 1 FN new_fuzzyController gt feedTable 1 1 EN new fuzzyController gt feedTable 2 1 FN new_fuzzyController gt feedTable 3 1 FN new_fuzzyController gt feedTable 4 1 FBN new fuzzyController gt feedTable 0 2 FP new fuzzyController gt feedTable 21 EP new fuzzyController gt feedTable 2 2 Z new fuzzyController gt feedTable 3 2 EN new fuzzyController gt feedTable 4 2 FN new fuzzyController gt feedTable 0 3 EP new_fuzzyController gt feedTable 3J EB new fuzzyController gt feedTable 2 3 FP new fuzzyController gt feedTable 3 3 Z new fuzzyController gt feedTable 4 3 2 new fuzzyController gt feedTable 0 4 FBP new fuzzyController gt feedTable 4 FBP new_fuzzyController gt feedTable
80. ong desktop home ilong fm gui THF MBE LV 4 997559 4 997559 5 000000 4 997559 4 997559 5 000000 root6ilong desktop home ilong 5 guit Figure 34 Data Collection Testing Results EJ Figure 35 Data Collection Testing Input Square Wave The software timer driven data collection code was able to accurately and timely collect voltage values ranging from 10V to 10V This is the same as the range of Kistler 5070A charge amplifier output voltages ensuring proper data reading As seen in the above Figure 33 a 5V to 5V 85Hz square wave was read and the output printed to the terminal every 100 milliseconds Code was written to implement the use of the PCM3718HO hardware timer to collect data The hardware interrupt driven code is located in Appendix C in the DataCollector c section For both timing methods examples from the Advantech user s manual were studied and modified with appropriate values 4 Although both software and hardware timer driven code were implemented only the software timer driven code was tested This was due to the hardware timer driven data collection code needing to be run on the board Two versions of the Data Collecting program were written The first uses a software timer QtTimer to poll the force acting on The Kistler 9129AA Multicomponent Dynamometer used to directly gather the tri axial forces acting upon the workpiece outputs electrical charges in three channels which correspo
81. ontrol input fuzzy sets had to be defined and output sets had to be tuned The two input sets were given the same parameters and one of the can be seen below in Figure 8 Ms pp Figure 10 Input Fuzzy Set The displayed range is 300N to 300N however the two membership functions on the end extend out to 10000N used to simulate infinity The three inner functions were each given a range of 300N with a standard isosceles shape This results in each function overlapping with the functions on either side of it Tuning the fuzzy controller involved altering the ranges and shapes of each output membership function again until an acceptable response was produced through simulation While it is important to be accurate with these numbers they do not have to be as exact as the gains in the PID because the fuzzy controller is much more forgiving in this sense Unlike the PID controllers Simulink was used in order to test the fuzzy controller The model used can be seen below in Figure 9 Constant1 Derivative 1 Controller with Ruleviewer Math Function Fuzzy Logic Controller with Ruleviewer1 Gain Figure 11 Fuzzy Control Testing Loop To understand the model it is easiest to start at the memory block This block is used to set the initial force for the test This sends a signal to three places a scope block to measure the signal a derivative block to calculate the change in force and the neg
82. or is cZ then Spindle Speed is sN 1 19 If Error is P and Change Error is cP then Spindle Speed is Z 1 20 If Error is P and Change Error is cBP then Spindle Speed is sP 1 21 If Error is BP and Change Error is cBN then Spindle Speed is sN 1 22 If Error is BP and Change Error is cN then Spindle Speed is sN 1 23 If Error is BP and Change Error is cZ then Spindle Speed is sN 1 24 If Error is BP and Change Error is cP then Spindle Speed is Z 1 25 If Error is BP and Change Error is cBP then Spindle Speed is sP 1 Figure 8 Fuzzy Logic Spindle Speed Rules 2 If Force i is BN po Detta Fecal is cN then Feed I Rate i is IN a 3 If Force is BN and Delta_Force is cZ then Feed_Rate is fP 1 4 If Force is BN and Delta Force is cP then Feed Rate is fP 1 5 If Force is BN and Detta Force is cBP then Feed Rate is fBP 1 6 If Force is N and Delta Force is cBN then Feed Rate is fN 1 7 If Force is N and Delta Force is cN then Feed Rate is fN 1 8 If Force is N and Delta Force is cZ then Feed Rate is fP 1 9 If Force is N and Delta Force is cP then Feed Rate is fP 1 10 If Force is N and Delta Force is cBP then Feed Rate is fBP 1 11 If Force is ZE and Delta Force is cBN then Feed Rate is fBN 1 12 If Force is ZE and Delta Force is cN then Feed Rate is fN 1 13 If Force is ZE and Delta Force is cZ th
83. p fkp cont gt feedKi fki cont gt feedKd fkd cont gt spindleKp skp cont gt spindleKi ski cont gt spindleKd skd cont gt errorBoundary errbound cont gt currFeedRate 0 cont gt currSpindleSpeed 0 dont touch theFuzz cont gt desiredSpindleSpeed 0 cont gt desiredFeedRate 0 void init as synchPID Controller cont float desF float fkp float fki float fkd float skp float ski float skd cont gt controllerType 0 synch printf Setting desF to f desF cont gt desiredCuttingForce desF cont gt feedKp fkp cont gt feedKi fki cont gt feedKd fkd cont gt spindleKp skp cont gt spindleKi ski cont gt spindleKd skd cont gt errorBoundary 0 cont gt currFeedRate 0 cont gt currSpindleSpeed 0 dont touch theFuzz cont gt desiredSpindleSpeed 0 cont gt desiredFeedRate 0 void init as Fuzzy Controller cont float eNeg float E float ePos float deNeg float DE float dePos float recFeed float recSpindle float Ks float u float doc cont gt controllerType 2 fuzzy cont gt desiredCuttingForce 0 actually we need to do the cutting force equations here cont gt feedKp 0 cont gt feedKi 0 cont gt feedKd 0 cont gt spindleKp 0 cont gt spindleKi 0 cont gt spindleKd 0 cont gt errorBoundary 0 cont gt currFeedRate 0 cont gt currSpindleSpeed 0 cont gt theFuzz defaultCreateFuzzyController eNeg E ePos deNeg DE dePos recFeed r
84. play plots of relevant information and display an alarm in the event of excessive tool wear breakage all via a user friendly interface Functionality to back the machining tool off of the workpiece within 10ms of extreme cutting tool abrasion tool breakage was not implemented due to time constraints and no available resources about communication protocol with the HNC210iB Although all of these functionalities were implemented few could be tested due to setbacks such as not receiving fully functional hardware until the week before the end of the project and not being able to obtain an optimal hard drive compact flash card Accordingly tasks whose outcome could be tested in simulation were reducing the possibility of errors occurring when the physical system is fully realized Even so a significant portion of all the necessary work to fully implement the system has been accomplished 6 Future Works 6 1 Response time As currently implemented the device runs on soft real time That is the 10ms response time to tool breakage is met most of the time however this is not guaranteed As the program currently stands it uses the Qt class Qtimer which has an accuracy on a Linux system such as this of 5ms 7 This is a similar problem for the 100ms response time for graphing and actively controlling the system In order to fix this issue hard real time must implemented using the hardware timer interrupts of the PCM3718HO Code
85. positive values float truths 5 An array representing the truths of each member function in the MFD mFD brief Create and initialize a MFD param i int functionality not implimented yet param BNeg memberFunction The leftmost member function param BNeg memberFunction The left member function param BNeg memberFunction The middle member function param BNeg memberFunction The right member function param BNeg memberFunction The rightmost member function return mFD The pointer to the new mFD NULL if fail if mFD createMFD int i memberFunction BNeg memberFunction Neg memberFunction Zer memberFunction Pos memberFunction BPos goes through each 5 membership diagrams and calls getTruth on them returns an array filled with the 5 truth values brief goes through each 5 membership diagrams and stores their truths for given input param mFD MFD The mFD in question param input float Input value to get truths for veturn void The truths are stored in mFD gt truths AY void getTruths mFD mFD float input brief gets the x axis centroid of the desired member param mFD MFD The mFD in question param i int Indicates which member function to get the centroid of Xreturn float The desired centroid Auf float getCentroidOfMember mFD mFD int i brief gets the area of the desired member param mFD MFD The
86. ptive control which are adaptive control with optimization ACO adaptive control with constraints ACC and geometric adaptive control GAC 1 ACO optimizes systems by analyzing a performance index subject to process constraints ACC systems don t rely on a performance index focusing solely on the constraints GACs maintain part quality through analyzing and compensating for tool wear and deflection Due to the project requirements and the provided hardware an ACC system was the chosen implementation As mentioned adaptive control with constraints maintains a given variable such as cutting force within provided constraints by altering cutting parameters such as feedrate and spindle speed The cutting parameters are manipulated through the use of some control algorithm that has been implemented such as a PID proportional integral derivative controller or a fuzzy logic controller Adaptive Control Systems for Machining describes a proportional controller in which feedrate is manipulated to maintain cutting force 1 In the article Fuzzy Control of Spindle Power in End Milling Processes a method of control using fuzzy logic to control both spindle speed and feedrate to regulate power is brought forth with reported success is various different scenarios 2 Design of a Drilling Torque Controller for a Machining Center describes a system in which a PID controller was used to control torque while drilling by manipulating the
87. r cont float mf f t gt numteeth s float force exp co gt Ks co gt DoC co gt u return 1 0 mf File IO class fileIO brief Functions for writing to file saving data author Brenden Gibbons Contact btgibbons wpi edu y allows use of this C file in CPP classes ifdef cplusplus extern C endif include guards ifndef FILEIO H define FILEIO H include queue h brief Outputs a float to a given FILE and inserts a new line character param stream FILE A pointer to the file to be written to MG int E int xf int id int pex param f float The float to print return int 1 if success 0 if fail outputFloatLine FILE stream float f brief Outputs a float to a given FILE then a tab character param stream FILE A pointer to the file to be written to param f float The float to print return int 1 if success 0 if fail outputFloatTab FILE stream float f brief Outputs a String to a given FILE then a tab character param stream FILE A pointer to the file to be written to param string char The string to print return int 1 if success 0 if fail outputStringTab FILE stream char string brief Outputs a String to a given FILE then a newline character param stream FILE A pointer to the file to be written to param string char The
88. reateMemberFunction 1 15 1 51 1 33 mf5 createMemberFunction 1 48 1 84 1 66 mFD3 createMFD 1 mfl mf2 mf3 mf4 mf5 output feed mfl createMemberFunction 15 51 33 mf2 createMemberFunction 48 84 66 mf3 createMemberFunction 82 1 18 1 mf4 createMemberFunction 1 15 1 51 1 33 mf5 createMemberFunction 1 48 1 84 1 66 mFD4 createMFD 1 mf1 mf2 mf3 mf4 mf5 fuzzyController the fuzzyController the fuzzyController createfuzzyController mFD1 mFD2 mFD3 mFD4 return the fuzzyController void Control fuzzyController FC float forcelIn temporary floats float error de get error and change in error error forceIn FC gt idealCuttingForce de error FC gt prevError update the truths for all member functions in the input MFDs getTruths FC gt errorInputMFD error getTruths FC gt deltaErrorInputMFD de feed in the truth arrays first errorinput then deltaerrorinput to applyFuzzyLogic applyFuzzyLogic FC FC gt errorInputMFD gt truths FC gt deltaErrorInputMFD gt truths set the error to the previous error FC gt prevError error float getNormalizedSpindleSpeed fuzzyController FC multiply the ideal spindle speed by the currentSpindleSpeed which is a percentage of ideal spindle speed and return it return FC gt idealSpindleSpeed FC gt currentSpindleSpeed float getNormalizedFeedRate fuzzyController FC multiply the ideal feed rate by the currentFeedRate
89. return 1 int writeQueueToFile FILE file Queue queue loop counter int i temporary float for i 0 i lt queue gt cells used i dequeue from the queue print the value to a file then re enqueue float tempfloat tempfloat float malloc sizeof float tempfloat float dequeue queue outputFloatLine file tempfloat enqueue queue void tempfloat I want to free it to avoid mem leak but there is a segfault free tempfloat fclose file int writeQueueToFileName char fname loop counter int i temporary float float tempfloat tempfloat float malloc sizeof float make new file Queue queue FILE file fopen fname w Open create file make sure it was created if file O for i 0 dequeue from the queue i lt queue gt cells used i tempfloat float dequeue queue outputFloatLine file tempfloat enqueue queue void tempfloat free the tempfloat free tempfloat fclose file return 1 failed return 0 int writeAllTheQueuesFormatted FILE file Queue queue3 Queue queued loop counter int i temporary float float tempfloat all queues will have the same number of entries outputStringTab file Time s outputStringTab file C Force outputStringTab file FeedRate outputStringTab file SSpeed outputStringTab file xFqueue outputStringT
90. ring the creation of the software to allow extra features or to ensure other ones were met Table 2 Requirements Analysis Requirement Customer e The system will be used in HNC s customer s factories on assorted milling machines in order to monitor tool wear and reduce tool breakage The system will be expected to operate effectively in a typical factory setting Architectural The system must be comprised of A usb mouse an analog keyboard a monitor an Advantech PCM3718HO an Advantech PCM3355 a Kistler 5070A charge amplifier and a Kistler 9129AA Multicomponent Dynamometer The NC model used is the HNC 210B Structural e Expandable Code base Object oriented style coding Data modules for easy data recording Behavioral Responds to Possible Cutting Tool breakage by setting off the Mill alarm showing a flashing red and black screen backing the milling tool away from the workpiece and stopping the spindle within 10ms Responds to Start and Stop buttons by turning on off data collection control and updated display of values Responds to pressing restart button in alarm screen by turning off the alarm saving all stored data and allowing new operations to commence Responds to clear button being pressed Allows the creation of Individual tools and operations Allows the creation of control systems Responds to Save Data button being pressed by writing stored data to a forma
91. s to comment all code consistently and thoroughly Each file in the code base is extensively commented with great detail in the DoxyComment format The final measure to ensure ease of expandability modification was to create a pdf file outlining how all code fits together and explaining each module by a line by line basis 3 4 Graphical User Interface Design As shown previously in the Requirements Analysis there were minimal specific requirements for the system s GUI The first step taken in creating a GUI was to choose the most appropriate and usable framework Possible frameworks included wx Widgets Qt GTK and BVRDE Ultimately QT was chosen due to its C cross platform nature signals and slots implementation and ease of use Additionally several team members had used it previously to avail The 2003 Qt version 3 1 0 was used to ensure that the PCM 3355 could run it in a Red Hat Linux 9 environment as required by HNC This version comes built in with larger RH9 installs so compatibility issues were reduced naturally In addition to providing a C programming framework the Qt install includes many useful tools such as Qmake for making large projects with ease and QML for documentation of both C and C GUI files Qwt widgets version 4 2 0 for Qt were used for displaying real time plots A process of programming a GUI analyzing it and iteratively improving it to better meet exceed requirements was used as a method to achieve
92. s were obtained Finally one week before the end of the project a PCM 3355 board was delivered in working condition The final hardware system diagram is shown in Figure 16 below HNC 210B Numerical Gul Advantech Kistler 9129AA Controller PCM3355 Dynamometer User Input Feed Rate Spindle Speed Force Data Electrical Charge User Input Force Data ADC Value Analog Keyboard Force Data Kistler 5070A Charge Analog Voltage Amplifier Advantech PCM3718HO Feed Rate Spindle Speed Figure 18 Hardware System Diagram The flow of data and in what form it is at any stage is clearly labeled The function of each component is explained in Table 3 below Table 3 Hardware Components Advantech PCM3355 Single board PC104 computer used for main control calculations driving the display handling user input and overall system coordination Advantech PCM3718HO Analog I O board used for both collecting analog workpiece force data from the Charge Amplifier and outputting digital words to the HNC 210B to change the feedrate and spindle speed LCD Display Displays the GUI allowing the user to interact with the program Analog keyboard and Input devices used to operate the system USB mouse HNC 210B The numerical controller model that the system is made to operate with Kistler 9129AA Multi Measures forces on the workpiece in the x y
93. t and allocating memory to hold the queue entries param max_cells Maximum entries in the queue return Pointer to newly allocated Queue structure Sr void Forml init Alarm Alarm Frame initialization alarmFlag 0 alarmCounter 0 alarmFrame gt hide gui stuff controllerTypes gt hide synchPIDFrame gt hide asynchPIDFrame gt hide fuzzyFrame gt hide init the controller theController create Controller Tools and Operations Inits newToolFrame gt hide toolIndex 0 maxToolsReachedLabel gt hide newOperationFrame gt hide operationIndex 0 maxOperationsReachedLabel gt hide Param table init the Ptable create paramTable initializing it NULL if error this will be used for the control and data saving speed dataCounterTop int timeStep 1000 10 dataCounter 0 create queues that will be used for graphs spindleQueue create queue dataSize forceQueue create queue dataSize feedQueue create queue dataSize create the controller and the datacollector theController init Controller theDataCollector createDataCollector dataSize 50 create the NcCommunicator the NcCommunicator create NcCommunicator MAKE THE FUZZY CONTROLLER memberFunction mfl memberFunction mf2 memberFunction mf3 memberFunction mf4 memberFunction mf5 mFD mFD1 mFD mFD2 mFD mFD3 mFD mFD4 1 inpu
94. t error mf1 createMemberFunction 130 70 100 mf2 createMemberFunction 80 20 50 mf3 createMemberFunction 30 30 0 mf4 createMemberFunction 20 80 50 mf5 createMemberFunction 70 130 100 mFD1 createMFD 0 mfl mf2 mf3 mf4 mf5 input delta error mfl createMemberFunction 125 75 100 mf2 createMemberFunction 75 25 50 mf3 createMemberFunction 35 35 0 mf4 createMemberFunction 25 75 50 mf5 createMemberFunction 75 125 100 mFD2 createMFD 0 mfl mf2 mf3 mf4 mf5 output spinlde mfl createMemberFunction 15 51 33 mf2 createMemberFunction 48 84 66 mf3 createMemberFunction 75 1 25 1 mf4 createMemberFunction 1 15 1 51 1 33 mf5 createMemberFunction 1 48 1 84 1 66 mFD3 createMFD 1 mf1 mf2 mf3 mf4 mf5 output feed mfl createMemberFunction 15 51 33 mf2 createMemberFunction 48 84 66 mf3 createMemberFunction 82 1 18 1 mf4 createMemberFunction 1 15 1 51 1 33 mf5 createMemberFunction 1 48 1 84 1 66 mFD4 createMFD 1 mf1 mf2 mf3 mf4 mf5 timerCounter timestep 1000 the_fuzzyController createfuzzyController mFD1 mFD2 mFD3 mFD4 create the time axis that will be used in all the plots 600 tenths of a second minute of data displayed also fill dummy array that will be used for plotting for int i 0 i lt dataSize i t i timeStep double i time axis yli 0 dummy array force i 0 feed i 0 spin
95. t of the fuzzy controller that is to say the inputs and the outputs Going along with the method described in Fuzzy Control of Spindle Power in End Milling Processes it was decided that the system should take in the current error and the change in change in force and output a new feedrate and spindle speed as opposed to a change in feedrate and spindle as done in the PID controllers 2 The next step was to create the two fuzzy sets to take in the error and change in error The authors used 7 membership functions per fuzzy set 2 Keeping the user in mind it was decided to reduce this number five to make it a bit simpler without losing too much functionality The membership functions were labeled as Big Negative Negative Zero Error Positive and Big Positive and shaped as triangles a standard membership function shape After this the output fuzzy sets were created again using five triangle membership functions and the same naming convention with the spindle or feed as prefixes for the spindle speed and feedrate outputs respectively Once all of the fuzzy sets were created a rule set needed to be created There needs to be a rule for each possible combination of input fuzzy sets that alters the feedrate and spindle speed in some way resulting in 25 different rules Again Fuzzy Control of Spindle Power in End Milling Processes was used as a starting place to create t
96. tEnabled true opselectlabel gt setEnabled true buttonsTabs gt setEnabled true opComboBox gt setEnabled true toolComboBox gt setEnabled true if not disable start button void Forml changeTool if toolComboBox gt count gt 0 sets the current tool to the corresponding one in the toolArray printf LOLOLOLOLOLOLOLOL Changing the currTool currTool toolArray toolComboBox gt currentltem make sure that there is a tool and an operation if opComboBox gt count gt 0 amp amp toolComboBox gt count gt 0 printf both an op and tool are present if there is no controller disable start button if confirmEntry the Ptable toolComboBox gt currentItem startButton gt setEnabled false else there is a controller so enable start button else startButton gt setEnabled true set theController to the one from the ParamTable clearOldData theController theController getAtIndex the Ptable gt currentItem else startButton gt setEnabled false void Forml changeOperation toolComboBox gt currentItem make sure that there is a tool and an operation if opComboBox gt count gt 0 amp amp toolComboBox gt count gt 0 printf both an op and tool are present if there is no controller disable start button if confirmEntry the Ptable toolComboBox gt currentItem startButton gt setEnabled false
97. tion return memberFunction The pointer to the new memberFunction NULL if fail EJ memberFunction createMemberFunction float left float right float center brief Get a truth value from a member function given an input value INPUT phase param mF memberFunction The member function in question pointer param input float The input value return float The truth value xy float getTruth memberFunction mF float input brief Gets area of the trapezoid formed by a line at y truth and the memberfunction OUTPUT PHASE param mF memberFunction The member function in question pointer param truth float The truth value veturn float The Area of the trapezoid R float getArea memberFunction mF float truth brief Gets centroid of the trapezoid formed by a line at y truth and the memberfunction OUTPUT PHASE param mF memberFunction The member function in question pointer return float The centroid on the x axis of the trapezoid EN special note if this is called before GetArea on a memberfunction midLeft and midRight will be set to zero so it will not find the centroid at all float getCentroid memberFunction mF endif ifdef _ cplusplus endif include lt stdlib h gt include lt stdio h gt include lt math h gt include MemberFunction h memberFunction createMemberFunction float left float right float center pointer to the new memberFu
98. tion h include Controller h typedef struct paramtable Controller table 25 25 unsigned short filltable 25 25 jparamTable paramTable create paramTable int addSet paramTable pt Controller newCont int toolnum int opnum Controller getAtIndex paramTable pt int toolnum int opnum int confirmEntry paramTable pt int toolnum int opnum fendif ifdef cplusplus tendif include lt stdlib h gt include lt string h gt include ParamTable h paramTable create paramTable paramTable new_paramTable new paramTable paramTable malloc sizeof paramTable init table to null Ut Ady a for ii Dr ii lt 25 dit for jj 0 3j lt 25 jj new_paramTable gt filltable ii jj 0 return new paramTable int addSet paramTable pt Controller newCont int toolnum int opnum pt gt table toolnum opnum newCont pt gt filltable toolnum opnum 1 return 1 Controller getAtIndex paramTable pt int toolnum int opnum printf getting at index d d n toolnum opnum return pt gt table toolnum opnum int confirmEntry paramTable pt int toolnum int opnum return pt gt filltable toolnum opnum Queue class Queue brief Queue used for all types of things author Brenden Gibbons Contact btgibbons wpi edu 7 allows the use of this C file in CPP classes ifdef cplusplus extern C fendif include
99. tted text file Quit button exits the program cleanly Two tabs for basic and advanced users Functional Monitors and display 3 axis force sensor data in real time 100ms versus time Determines the state of tool breakage at any given moment Display real time 100ms plots of the current feedrate and spindle speed versus time Display numbers correlating to values of plots in real time Actively control the feedrate and spindle speed of the NC to extend tool life using a selected control algorithm Non Functional Usable by basic Factory worker Robust and hard to break extensively tested Efficient enough to run on the limited hardware selected Performance 10ms hard real time response for setting off the alarm and backing the mill of the workpiece in the event of likely tool wear 100ms soft real time response for collecting data controlling the feedrate and spindle speed and displaying collected data Design Whole program done in C except for GUI elements GUI elements done in C Use Red Hat 9 Linux as the operating system Expandable code base In order to best meet all of the requirements of the preceding analysis several features of the code were planned out in detail before any coding began It was decided that for ease of usability compatibility and portability that all non C code would be styled in an object oriented manner This involved the extensive use of structs and passing pointers to th
100. turn the magnitude of the latest entries in a dataCollector s Queues param the dataCollector dataCollector the dataCollector to act on return float The magnitude of the latest Queue entries SA float forceFloat dataCollector the dataCollector float getCurrForceSquared dataCollector the dataCollector float getForceFromFS dataCollector the dataCollector Tool t cuttingOperation co Controller cont endif ifdef cplusplus fendif include lt stdlib h gt include lt stdio h gt include lt sys types h gt include lt sys stat h gt include lt fcntl h gt include lt unistd h gt include lt sys ioctl h gt include lt string h gt include lt sys mman h gt include lt termios h gt include lt signal h gt include lt Advantech advdevice h gt GA include queue h include fileIO h include DataCollector h define MODE 0 dataCollector createDataCollector int size int srate int m dataCollector new dataCollector pointer to the new dataCollector allocate mem for datacollector new dataCollector dataCollector malloc sizeof dataCollector if new dataCollector NULL return NULL Error unable to allocate fill in the struct set the sample rate new dataCollector gt samplingRateHz srate printf The sampling rate is d n srate printf The sampling rateHZ is d n new dataCollector gt samplingRateHz set the mode new dataColl
101. ue which queue ptr void dequeue Queue which queue Check if queue is empty if which queue gt cells used lt 0 which_queue gt cells_used 0 insurance return NULL Queue empty check if head is running off the end if which_queue gt head gt which queue gt end wrap head around to beg which queue gt head which queue gt beg dequeue void R which queue gt head remove the head pointer which queue gt head Remember this points which queue gt cells used decrement the cellsused return R return the head pointer void printQueue Queue which queue int i loop counter loop to print the queue for i 0 i lt which queue gt cells used i pointer to a float float tempfloat This will only print the queue if it has floats tempfloat float malloc sizeof float dequeue one from queue tempfloat float dequeue which_queue print the value printf Queue Entry Sd Sf i tempfloat reenqueue enqueue which_queue void tempfloat return void peek Queue which queue check if empty if which queue gt cells used lt 0 which queue gt cells used 0 insurance return NULL Queue empty void ret which queue gt head return ret void peekTail Queue which queue check if empty if which queue gt cells used lt 0 which queue gt cells used 0
102. wCont closeAsynchPID enable the start button startButton gt setEnabled true void Forml closeAsynchPID newAsynchRecForceEdit gt clear newAsynchfKpEdit gt clear newAsynchfKdEdit gt clear newAsynchfKiEdit gt clear newAsynchsKpEdit gt clear newAsynchsKdEdit gt clear X 7 newAsynchsKiEdit gt clear AsynchCutoffForce gt clea asynchPIDFrame gt hide enableBackground OG void Forml popFuzzyFrame controllerTypes gt hide fuzzyFrame gt raise fuzzyFrame gt show fuzzyFrame gt repaint void Forml saveFuzzy make a new controller and pointer to the controller Controller newCont create Controller init the controller as an AsynchPID with inputted params init_as_Fuzzy newCont fuzzyENegEdit gt text toFloat fuzzyEZeroEdit gt text toFloat fuzzyEPosEdit gt text toFloat fuzzyDENegEdit gt text toFloat fuzzyDEZeroEdit gt text toFloat fuzzyDEPosEdit gt text toFloat fuzzyRecFeedEdit gt text toFloat fuzzyRecSpindleEdit gt text toFloat Di now save the controller to the Ptable addSet the Ptable newCont toolComboBox gt currentItem opComboBox gt currentItem closeFuzzy clearOldData theController theController newCont enable the start button startButton gt setEnabled true void Forml closeFuzzy fuzzyENegEdit gt clear fuzzyEZeroEdit gt clear fuzzyEPosE
103. were taken The raw output voltages of the charge amplifier range from 10V to 10V The Advantech PCM3718 board s Analog to Digital Converter is twelve bits number of states is 4096 This gives a resolution of 00488V bit The output from the ADC is therefore multiplied by this value to obtain the actual voltage Once an actual voltage has been obtained this voltage can be converted into a force value by first subtracting ten volts to normalize the range the multiplying by 10000Newtons 20V A function for this purpose can be found in Appendix C DataCollector c Two methods for collecting data were implemented The first using a software timer to poll the charge amplifier is an easy to implement yet somewhat unreliable method It involves simply using the advDAQ method to get an ADC value from a channel on a software driven interrupt Problems can occur however when other software processes are scheduled to be executed before this timer interrupt happens and therefore it may not be accurate every time The second method is using a hardware interrupt This is slightly more difficult to implement than the previous method however it is more precise and reliable A hardware interrupt is set on the PCM3718 to trigger every so often Then when this occurs an interrupt service routine is run to record the data Because the timer on the PCM3718 is very reliable and the interrupt service routine always pre empts the currently running process it is rar
104. y Force Queue Queue zForceQueue pointer to the z Force Queue int lastvalue used for simulation int cf used for simulation int cs used for simulation dataCollector brief Create and initialize a DataCollector param size int The size of the three queues param srate int The sampling rate of data acq param mode int 1 if simulation 0 if real return dataCollector The pointer to the new dataCollector NULL if fail AY dataCollector createDataCollector int size int srate int m brief Make a random int param high int The highest random value to receive Xreturn int Pseudorandom number between 0 and high Kf int makerand int high brief Get an analog voltage value off of a specified port param channelNum int The port number in question param m int 1 if simulation 0 if run on board Xreturn float representing the AD value we int pullAnalogPort dataCollector the dataCollector int channelNum int m brief Convert a raw AD input to a force param rawInput float The raw AD value return float number representing the axial force ay float toForce int rawInput brief Get and store the forces on all 3 axes in their respective queues param the dataCollector dataCollector The dataCollector in question return int 1 for success 0 for failure int getAndStoreForces dataCollector the dataCollector brief re
105. zetatt A dataCounter 0 void Forml test void Forml unAlarmGUI turn off alarmflag nide the alarm frame alarmFrame gt hide alarmFlag 0 stop reset void Forml alarmGUI killTimer tehtimer perform the alarmRoutine aka spindle off back off the piece alarmRoutine the NcCommunicator now display the alarm in GUI alarmFrame gt raise alarmFrame gt show already a timer in use so this works well alarmFlag 1 save the Data to provide insight as to what went wrong saveData void Forml popNewToolFrame cancelNewOperation newToolFrame gt setIsTopLevel true newToolFrame gt resize newToolFrame gt raise newToolFrame gt show newToolFrame gt repaint disableBackground void Forml popNewControllerSelect need to ensure that both a tool and an operation are available if toolComboBox gt count lt 0 TODO pop up a warning that a tool must be selected printf A tool must be selected n else if opComboBox gt count lt 0 TODO pop up a warning that an operation must be selected printf an op must be selected n need to ensure that for the specified combo a Controller doesnt already exist else if confirmEntry the Ptable toolComboBox gt currentItem opComboBox gt currentItem 0 controllerTypes gt raise controllerTypes gt show controllerTypes gt repaint disableBac
106. zy Button is pressed Once all of the parameters for creating a new Fuzzy Controller have been entered the user clicks the Create button A fuzzy controller with two member function diagrams of five member functions each is dynamically created from the entered values Create New Fuzzy Controller Error Negative Error Zero Error Positive Change in Error Neg Change in Error Zero Change in Error Pos Specific Cutting Force U Depth of Cut Rec Feed Rate Rec Spindle Speed Create Cancel Figure 30 GUI Fuzzy Controller Setup The Asynchronous PID button opens the Create new Asynchronous PID Controller dialog Figure 29 Here the user inputs all of the parameters necessary for creating an Asynchronous PID Controller Once Create is clicked the controller is created and bound to the specific Tool Operation combination Create New Asynchronous PID Controller Recommended Force N Feed Kp Feed Kd Feed Ki Spindle Kp Spindle Kd Spindle Ki Cutoff Force N Figure 31 GUI Asynchronous PID Controller Setup The Create new Synchronous PID Controller dialog Figure 30 opened with the Synchronous PID button is identical to the Create new Asynchronous PID Controller dialog explained previously except that it lacks an input box for Cutoff Force as it is not needed for a synchronous PID contro

Download Pdf Manuals

Related Search

Related Contents

CASHLITE 2010 - Benutzerhandbuch © DATEX GmbH  着ぐるみ使用マニュアル  イモビハ−ネス配線全体図 裏面 に続く  BearTracks User Manual  Engineer-To-Engineer Note EE  How to disassemble a SX  SH Locator - Geotronics  MetScape 2.3.2 User Manual  Siig CE-MT1F12-S1 flat panel wall mount  Manual Exaustores - Tradicional - 350.cdr  

Copyright © All rights reserved. Failed to retrieve file