Home

Icon - To Daim

1. 20 CHAPTER 2 THE INFRASTRUCTURE Line Of Sight The Line Of Sight module s main function is to generate the 3DOF reference vector x It takes in the outputs from the previous modules in addtion to the lookahead distance A tuning parameter for s gradient algorithm u and the virtual point mass coordiantes in O frame q Equation 6 to 8 40 to 46 and 50 to 70 in Skjetne 2014 are implemented here It s outputs are x dynamic LOS assigment for d f and dynamic LOS assigment for fs 2 6 4 Control The Control module s main function is to generate a 3DOF command output 7 and a set of thruster commands Te It is currently made up of five modules one switch one thruster allocation and three control module It is set up to handle three controllers but can be expanded by adding and increasing the number of ports in the switch 0 a 1 A 2 Control Switch Thruster allocation Figure 2 8 Control module in the Simulink diagram 2 6 SIMULINK 21 Control module input controlModeSelector ctrlReset 80 qo Td ASry ASpx ASry ASnx L2 R2 Bow Thruster BT power VSP speed Parameter for switching between controllers Parameter for reseting controllers Initial value of s Initial values of q Desired position and orientation in O frame Position and orientation in Q frame Velocities in 5 frame 3DOF reference vector in O frame col q 10
2. 51 Force N Smoothing of f 4 08 06 04 02 0 02 04 06 08 u value setting Figure A 6 Measurments of VSP speed at 0 4 coefficient of CSE1 52 APPENDIX A THRUSTER PLOTS Force N Smoothing of f smoothing 4 0 8 0 6 0 4 02 0 02 O04 06 08 1 u value setting Figure A 7 Measurments of VSP speed at 0 4 coefficient of CSE1 53 Force N Smoothing of f smoothing 4 0 8 06 04 0 2 0 02 04 O06 08 u value setting Figure A 8 Measurments of VSP speed at 0 4 coefficient of CSE1 54 APPENDIX A THRUSTER PLOTS Force N Measured f and lookup table f L Measured f g Lookup table f 4 0 8 06 0 4 02 0 02 O04 O06 08 1 u value setting Figure A 9 Measurments of VSP speed at 0 4 coefficient of CSE1 55 Force N Measured and lookup table f L Measured Lookup table t 4 0 8 06 0 4 02 0 02 04 O06 08 u value setting Figure A 10 Measurments of VSP speed at 0 4 coefficient of CSE1 56 APPENDIX A THRUSTER PLOTS Force N Measured f and lookup table f L Measured f Lookup table f 4 0 8 06 O04 02 0 02 04 O06 08 1 u value setting Figure A 11 Measurments of VSP speed at 0 4 coefficient of CSE1 57 Force N Measured f and lookup table f L Measured f Lookup table f 4 0 8 06 04 0 2 0 02 04 O06 08 u value
3. This report tries to cover the main points of the original an overview of the infrastructure used by CyberShip Enterprise 1 the work related to it and laboratory experiment carried out with it I want to first and foremost thank my supervisor Roger Skjetne for his advice knowledge and suggetions Without those the final control architecture would be a real mess to work with And also for the opportunity to with CyberShip Enterprise 1 I would like to thank my co advisor ivind K Kjerstad for his inputs and support through out this endeavor expecially durring the the frustrating periods of debugging I would also like to express my graditude to Senior Engineer Torgeir Wahl for his assistance when working in the laboratory for teaching me how to utilize the various equipments and for developing the software need to improve the overall reliability Lastly a thanks to my familiy for nugging me in this direction Looking back working on this pre report writing have been the highlight of my life up to this point I have learned more the last year than any of the previous ones vi Preface Contents Project description Summary Preface 1 Introduction Ld 1 2 1 3 1 4 1 5 1 6 1 7 1 8 2 The 2 1 232 2 9 2 4 2 5 2 6 MOVANO ELT T Bookera 4 4 duos usce desee o eg e NR ae PLACA Previous works i sia e saoe dw wacka wok modem de ok eed we amp 2320s ueque Ae e
4. Partial differentiation of x Partial differentiation of x Partial differentiation of Vios Partial differentiation of Vios Partial differentiation of Vros Partial differentiation of Vios Partial differentiation of Vios Dynamic LOS assigment for q Partial differentiation of f Partial differentiation of f Partial differentiation of f Dynamic LOS assigment for Partial differentiation of fs Partial differentiation of fs Partial differentiation of fs Inertia matrix Hydrodynamic damping matrix Thruster configuration matrix Diagonal tuning matrix Diagonal tuning matrix Diagonal tuning matrix Diagonal tuning matrix Tuning parameter for virtual control a Gradient update law tuning parameter Up Down position of Left Analogstick Left Right position of Left Analogstick Up Down position of Right Analogstick Left Right position of Right Analogstick Shoulder botton signal Shoulder botton signal BT power limit VSP speed setpoint 22 CHAPTER 2 THE INFRASTRUCTURE Control module output T Force vector in B frame Te Thruster commands set s Path parameter of desired path p q m Virtual point mass coordiantes of vessel in O frame Control Within each Control n different kind of control design can be implemented for mod ularity and structure one per subsystem However they can be placed anywhere as long as it uses a GoTo block to declare the 7 produced as a global v
5. is connected to HILlab the Main battery large fat one is above 12 Volt the Servo battery small slim one is above 6 Volt Place Main battery large fat one beneath wireless anntenna adjacent to waterproof box between the wires with battery terminals furthest away from it Place Servo battery small slim one at bow between tunnel thruster and waterproof box with battery terminals closes to the waterproof box Postive battery terminal RED port at portside and negative battery terminal BLACK port at starboard side Connect wire with red isolation RED wire to RED port and wire with black isolation BLACK wire to BLACK port Connect first the RED wire before the BLACK wire to the batteries The Main battery large fat one should be connected first then wait a few sec 5s before connecting the Servo battery small slim one Note it should not matter in which order it is done but from experience connectiong RED wire before 73 74 APPENDIX B DRAFT FOR USER MANUAL BLACK wire gives a much higher probability for communication with the CompactRIO on Cybership Enterprise 1 99 1004 ish than connecting the BLACK wire before the RED wire 25 ish and it is a habit to connect main before the servo since main powers CompactRIO while servo powers D Link wireless bridge There should be 3 red lights lighting up one at bow in a purple box for indicatiing power to tunnel
6. CSE1 65 Force N VSP speed 0 2 08 0 6 04 02 0 02 04 06 08 Uy Figure A 20 Measurments of VSP speed at 0 2 coefficient of CSE1 66 APPENDIX A THRUSTER PLOTS Force N VSP speed 0 3 us Figure A 21 Measurments of VSP speed at 0 2 coefficient of CSE1 67 Force N VSP speed 0 2 0 2 0 15 0 1 0 05 1 08 0 6 04 02 0 02 04 06 06 1 U5 Figure A 22 Measurments of VSP speed at 0 2 coefficient of CSE1 68 APPENDIX A THRUSTER PLOTS Force N VSP speed 0 2 kar 08 0 6 04 02 0 02 04 06 08 1 Us Figure A 23 Measurments of VSP speed at 0 2 coefficient of CSE1 69 Force N BT power 0 15 Us Figure A 24 Measurments of BT power at 0 15 coefficient of CSE1 70 APPENDIX A THRUSTER PLOTS Force N BT power 0 30 0 2 0 15 0 1 0 05 0 0 05 0 1 0 15 0 2 Us Figure A 25 Measurments of BT power at 0 3 coefficient of CSE1 71 Force N BT power 0 40 L us Surge us Sway 0 5 1 5 0 25 02 015 01 005 0 005 01 O16 02 025 Us Figure A 26 Measurments of BT power at 0 4 coefficient of CSE1 72 APPENDIX A THRUSTER PLOTS Appendix B Draft for User Manual Connecting to Cybership Enterprise 1 RT CompactRIO NI cRI09024 CSE1 192 168 0 77 Make sure the ethernet cable is connected to ACT LiNK port 1 and to the D Link Wireless Bridge the Laptop
7. CSE1 and Qualisys togehter It connects to mdl file or the derived files dll vxworks in Model and Host In Mappings the connection between the blocks in Block Diagram and the blocks in mdl file is established The relationship of the SIT output block in mdl with CSE1 the thrusters through cRIO are set in Hardware I O in connection with the FPGA bitfile As well as the SIT input block in mdl for the battery voltages and any force ring connected to the cRIO The link between Qualisys and the re maining SIT input block in is done indirectily through IO llb and Base Rate Loop vi The is automatically created by SIT It contains six vi files Base Rate Loop vi Close Init Read Write vi and Ref ctl It is in Base Rate Loop vi the whole Qualisys data acquisition is handled The others are called but it is not worth going into detail Base Rate Loop vi makes use of a driver vi created by Senior Engineer Torgeir Wahl 14 CHAPTER 2 THE INFRASTRUCTURE 2 5 5 Qualisys Track Manager Drivers The original driver used in Skatun 2011 also created by Torgeir Wahl aquired processed and sent the data all in the same timestep The consequence of this forced the time step of the mdl to be the same as the sample rate of Qualisys If Qualisys had a higher sample rate than the rate the mdl file was solved the mdl file would would constantly be working with older and older data as the time progressed If t
8. and output 7 and v It consists of Input from SIT and Navigation Switch Input from SIT Navigation Switch Figure 2 10 Navigation module in the Simulink diagram Input from SIT This subsystem have remained mostly the same since the orginal subsytem found in Skatun 2011 The major difference is the passive low speed observer used to estimate the velocities instead Its function is to process the Qualisys values received through the SIT server and output nos and vos In addtion it needs Te M D for the observer It also sends out other parameters given by Qualisys seperatly battery voltages The passive low speed observer was introduced into the stucture by Co advisor ivind Kjerstad due to the noisy velocities created when using a Derivation block It is a modified version of an observer from MSS GNC Toolbox 2 6 SIMULINK 25 Navigation Switch The Navigation Switch is similar to Control Switch it uses a varable called con trolInputSelector to decide if it shall send the valuse from Qualisys or the simulator The addtional inputs are qrs 705 VLS and vos The outputs are 7 and v 2 6 7 C S Enterprise 1 Matrices The function of this subsystem is to define CSE1 s matrices it is created to make it easy and efficient to modify without having to check if all the places the matrices are used are up to date It does not have any inputs but it is possible to directly map controls from LabVIE
9. are used for compact structuring The right half are different types of visualiza tion of the process taking place in the background a 3D visualization and plots of key variables The various controls indicators and plots should be self explanatory based on the name label 2 5 3 Block Diagram This is where the each element in the Front Panel is defined with repect to interaction behavior and data to display among one another In general it is here the mapping are done However since the model structure and dynamics are created in Simulink the true mapping happens in SIT which in turn automatically create the mapping in the Block Diagram The diagram is basically the same as Sk tun s It have been organized and tweaked for relative path definition and with added comments and lables It can be divided into four 2 5 LABVIEW 13 main groups First is the loop stucture that handles the signals from the PS controller via BTSix and PPJoy Second is the blocks used for the 3D visualization The third group is the stuctures created by SIT The last group is the miscellaneous group scattered all across the diagram containing tab control and unmapped blocks Unless they are wired to anything they can be more or less freely be placed anywhere in the diagram Figure 2 4 Overview of LabVIEW Block Diagram 2 5 4 Simulation interface toolkit SIT is the intersection that connects LabVIEW Block Diagram and Base_Rate_Loop vi Simulink
10. are within the subsystems located in Main Subsystems Plant CSE1 ac tuator The block are color coded where green means Source red means Sink orange means GoTo and magenta means From The solver used is ode5 Dormand Prince with 0 1 as fixed step size Other solvers can also be used it depends on the complexity of the system and if the solver is able to finish within the time step When compiling the fiile using Real Time Workshop the only way to set the frequency of the model is by choosing fixed step If variable step is chosen then Real Time Workshop will decide the frequency A small sidenote Some comments and names of blocks in the mdl files may not be up to date for what ithey are actually used for 16 CHAPTER 2 THE INFRASTRUCTURE SignalProbe Inputs from LabVIEW Main Subsystems Outputs to LabVIEW Figure 2 6 Top level in Simulink diagram 2 6 1 Input from LabVIEW The function of this subsystem is to gather all input mappings from LabVIEW in one place It consists of Constant blocks to map to and GoTo blocks to declare them as global variables The signals mapped here are structured into scalar vector or matrix depending on the application of the signal before being declared a global variable 2 6 2 Output to LabVIEW This subsystems functions is similar to the Input from LabVIEW it gathers all mapping in one place just for outputs instead It consists of Fro
11. experimental results Tentative 8 Design implement and test an underactuated LOS maneuvering control law for CSEI Guidelines The scope of work may prove to be larger than initially anticipated By the approval from the supervisor described topics may be deleted or reduced in extent without consequences with regard to grading The candidate shall present his personal contribution to the resolution of problems within the scope of work Theories and conclusions should be based on mathematical derivations and logic reasoning identifying the various steps in the deduction The report shall be organized in a rational manner to give a clear exposition of results assessments and conclusions The text should be brief and to the point with a clear language The report shall be written in English preferably US and contain the following elements Abstract acknowledgements table of contents main body conclusions with recommendations for further work list of symbols and acronyms references and optionally appendices All figures tables and equations shall be numerated The original contribution of the candidate and material taken from other sources shall be clearly identified Work from other sources shall be properly acknowledged using quotations and a Harvard citation style e g natbib Latex package The work is expected to be conducted in an honest and ethical manner without any sort of plagiarism and misconduct Such practice is taken v
12. minimum to manually control CSE1 through a PS controller To run it requires LabVIEW with SIT package BTSix and PPJoy installed and the vi file the mdl file or it s derived files a wirelss network a PS controller with Bluetooth and Dongle In theory this enables CSE1 to not be confined to only the MC Lab Thruster HMI vi was deveopled for the purpose of measuring the thruster forces pro duced The input values needed to be fixed over a period It is an expansion of PS3 HMI vi allowing direct thruster control from the Front Panel StudentHMI vi is one of the last version of the GUI and the one who is most similar to Skatun s program of the four It can also be viewed as a simplified version of Template HMI vi It was created for the avarage student to easily use and modifiy It contains a manual PS thruster control a Dynamic Positioning DP control with setpoint and a Control Lyapunov Function LgV control with linear and ellipse path Template HMI vi is the final product and will be the main focus It is a mostly generic GUI with a setup for manual PS thruster control two automated control systems with the option of linear or ellipse path or setpoint It is in essence the same as Sk tun s program see Skatun 2011 for a simple introduction It requires data from qualisys to make use of the automated control systems 12 CHAPTER 2 THE INFRASTRUCTURE 2 5 2 Front Panel This is where the operator
13. of nine markers With the marker at bow bent forward and the marker at stern bent in the opposite direction The stand was removed since five markers gives a redundancy of two markers and the stand could be used for a seperate body The marker at bow was later straighten to be as vertical as possible due to constant shift of it s position caused by CSE1 crashing bow first into the basin walls 2 2 3 Thrusters The thrusters receives control signals from cRIO The bow thruster is a simple tunnel thruster driven and controlled by a hobby motor Each VSP have a motor for rotation and two servos to position the stick The distance between the stick and servos are fixed However they move in an circular motion which needs to be accounted for 2 3 Marine Cybernetics Laboratory The basin s dimensions imposes constraints on the maneuvering space of CSE1 As stated in Norwegian University of Science and Technology n d the basin have a total length of 40 meters a width of 6 45 meters and it can be filled up to a depth of 1 5 meters However measuring it from end to end it is closer to 39 meters and the length the real length where a vessels position is measurable is limited by several factors 10 CHAPTER 2 THE INFRASTRUCTURE The basin s water in and outlet are located at the far end of the basin aswell as it s beach On the opposite is the wave maker Between those are the measuring equipment mounted on the front side of the towing ca
14. possible to give a clearer understanding It is assumed that the reader have read or have access to Skjetne 2014 and Skatun 2011 Their contents are omitted but referred or asssumed known to the reader 1 8 Structure The thesis is structured into five chapters Introduction Infrastructure System identifica tion Control experiments and Conclusion The introduction have already been covered In the Infrastructue chapter a general overview is first presented followed by a description of each component with comments The System identification chapter documents the various procedures conducted in the MC Lab for determining the coefficients of CSE1 s hull and the thrusters In Control experiments the two controls are presented analysed and simulated and the laboratory run results are disscused In the last chapter the main points are summarized and suggestions for future work are presented based on those CHAPTER 1 INTRODUCTION Chapter 2 The Infrastructure This chapter presents the various components and equipments and how they interact with each other In the first section an overview of the infrastructure is presented The following sections provides a more detailed view of each component of the infrastructure 2 1 Overview server i i i i PlayStation Operator Donge E Mk controller Bluethoofh E a mta Wireless Network Dy j Wreles
15. setting Figure A 12 Measurments of VSP speed at 0 4 coefficient of CSE1 58 APPENDIX A THRUSTER PLOTS Force N Measured and lookup table fs Measured fz Lookup table fz 4 0 8 06 04 02 0 02 04 O06 08 1 u value setting Figure A 13 Measurments of VSP speed at 0 4 coefficient of CSE1 59 Force N Three approach to VSP surge force S Superposition of f f amp Simultaneous Superposition of lookup table f and f 0 4 0 2 0 02 04 06 08 1 u value setting Figure A 14 Measurments of VSP speed at 0 4 coefficient of CSE1 60 APPENDIX A THRUSTER PLOTS Three approach to VSP sway force Force N S Superposition off f 4 amp Simultaneous tong Superposition of lookup table and f 41 0 8 06 04 02 0 02 04 O06 08 1 u value setting Figure A 15 Measurments of VSP speed at 0 4 coefficient of CSE1 61 Force N VSP speed 0 3 Uy Figure A 16 Measurments of VSP speed at 0 3 coefficient of CSE1 62 APPENDIX A THRUSTER PLOTS Force N VSP speed 0 3 us Figure A 17 Measurments of VSP speed at 0 3 coefficient of CSE1 63 Force N VSP speed 0 3 Us Figure A 18 Measurments of VSP speed at 0 3 coefficient of CSE1 64 APPENDIX A THRUSTER PLOTS Force N VSP speed 0 3 Us Figure A 19 Measurments of VSP speed at 0 3 coefficient of
16. the circular motion of the servos Some of it can be caused by the rotation when measureing or due to placement of thuster However this can not fully explain the notable force meaured in the other direction Of the servos servo 4 have the strongest coupling in surge sway Two advance control design were implemeted tuned and tested a LgV backstepping and a Nonlinear PID designed controller In the ideal simulated word both of them were equal in terms of maneuvering Only ellipse path was used in the laboratory This had to do with space constraints and a wish to have long run time on the experiments 35 36 CHAPTER 5 CONCLUSION Originally only the LgV backstepping was tested in the laboratory It converged and per formed well The only downside was it would constantly overshoot the heading resulting in a constant oscillation while moving alone the path The reason was due to the noisy velocity estimation Other opportunity presented itself to run more experiments in the laboratory However unknown at that time and discoved to late one of the servo arms broke due to fatigue from the constant oscillation of the previous runs corruping the mapping and observer estimates In the laboratory their performance depended heavily on how they were imple mented If both of them were fully implemented LgV backstepping proved to handle the uncertainties better Although it would overshoot when exiting the sharper part of the ellipse path The
17. to die out and reset the zero settings The first measurment set did not have a zero measurment before the distubance is instroduced e g towing or activating CSE1 which was introduced in the latest measurment set The first set of measurments was done for towing the hull 0 45 and 90 degrees VSP force generation with a VSP speed set to 0 4 with 0 1 step size and for BT with power limit at 0 5 also 0 1 For the thrusters the input value ranged from 1 to 1 The second set were done just for the thrusters for lower speed and power VSP speed for 0 2 and 0 3 and BT power at 0 15 0 3 0 4 and 0 5 Figure 3 1 The setup of CSE1 for system identification 27 28 CHAPTER 3 SYSTEM IDENTIFICATION 3 1 Hull The measurements from 0 and 90 degrees towing yielded Force acting in surge direction on Cybership Enterprise from towing Measured dampning force Fitted curve XU Kuju uu XU Fitted curve X u DP X 7 014492 Xu 3 10143 Xu 0 34354 X 0 59739 Mean drag force N 0 8 0 6 0 4 0 2 0 2 0 4 0 6 Surge speed m s Figure 3 2 Measurments of dampning coefficient of CSE1 The values for X and Y are similar to Skatun s values And a new value can be added for slow speed N 0 18140 3 2 Thruster mapping The VSP have been measured for 0 2 0 3 and 0 4 in VSP speed while the BT have been measured for 0 15 0 3 0 4 and 0 5 in power limit The starboard voith schneider prop
18. vessel is able to follow the virtul point it is chasing but the virtual point is unable to stick to the desired path While the fully implemented Nonlinear PID would struggle to converge to the path and behave erratic The velocity dependent term distorted Nonlinear PID s results the most The reason is most likely due to the quality of the estimations of the velocities from the observer The feedfoward term also deteriorate the control and prevents it from converging If the Nonlinear PID was partially implemented deactivating the feedforward and the velocity dependent term then it is the superior one It would converge naturally to the desired position Once on it would stay there indefinitely despite of the broken servo arm 5 1 Future works With all the improvement in the reliability one important problem still remains the loss of visibility There are incidents when Qualinsys displays it sees all markes but it is unable to calculate the vessels position and orientation The cause of this needs to be further investigated Spare part servo arms should be purchased and padding for the hull to dampen the impact to the basin walls Several crack have been observed on the hull Simulation Interface Toolkit have been discontinued modifying it for Modular Interface Toolkit can be an option The labeling and choice of variable name can be improved upon There is a greate improvement potential of the contents of this report Graphic
19. 2 2g y ryk cos ks 2 2h 2 21 where xp and yo are coordinates of origin of ellipse path in Q frame r and r are radius of ellipse path and is a scaling parameter of path parameter s often much smaller than 1 The values are merged together into p col x y for each of their respective partial differentiations and the module implements equation 36 in Skjetne 2014 The workaround module is used due to unknown technical limitations that wont running or compiling when a regular Switch block for switching between paths It uses a variable dubbed pathSelector that can be either 1 or 0 The workaround make use of Pa paypathSelector pyo 1 pathSelector 2 3 For each partial differentiations where p is the chosen path is the linear path and Pao is the ellipse path If pathSelector is 0 then ellipse path is chosen If it is 1 then linear path is chosen Heading The Heading module make use of the outputs from the Path module It implements equation 2 and 47 to 49 in Skjetne 2014 and outputs the desired heading v and it s partial differentiations 74 and ye Speed Assigment The Speed Assigment module require the outputs from the Path module in addtion to the desired forward speed ug It implements eappropriatequation 5 38 and 39 in Skjetne 2014 and passes on the speed assigment vs it s partial differentiations and the time derivative of ug
20. 8 11 13 22 39 QTM Qualisys Track Manager 4 10 SIT Simulation Interface Toolkit 4 8 10 11 13 15 24 39 VSP Voith Schneider Propeller 8 9 21 22 24 27 28 44 41 42 Acronyms Symbols Virtual controller 21 43 T Gradient update law tuning matrix 21 22 A Lookahead distance m 17 20 Position and heading in O frame vector col p 3 21 25 33 na Desired position and heading in O frame vector col p v4 33 k Tuning parameter for virtual control o 21 22 A Gradient update law tuning parameter 21 22 u Tuning parameter for s gradient algorithm 17 20 v Velocity in B frame vector col u v r 3 21 25 T Contol output col X Y N 3 20 22 24 x 3DOF reference vector in Q frame 18 20 21 33 w Heading in O frame in radians rad 3 43 44 Desired heading in O frame 19 LOS heading in Q frame 18 21 B Thruster configuration matrix 3 21 22 25 D hydrodynamic damping matrix 3 21 25 fac Thruster force vector col fi f2 fs fa fs 3 4 22 f Dynamic LOS assigment for 18 20 21 fs Dynamic LOS assigment for s 18 20 21 k Scaling parameter of path parameter s 17 19 Kp Diagonal tuning matrix 21 22 K Diagonal tuning matrix 21 22 K p Diagonal tuning matrix 21 22 43 44 Symbols M Inertial matrix 3 21 25 N Moment about z axis Nm 43 p Vessels position as a point mass in O frame vector col x y 3 19 43 p Desired position al
21. Connection Manager ZI 1 Hardware I O gt Configure HW I O 2 Easiest with Import and select the hardwaremapping file or Right click on Device Tree on the IP address 192 168 0 77 3 Select Add Device gt NI FPGA 4 In NI FPGA Property Dialog The FPGA Target should be most often something ending with cRIO 9113 The FPGA Bitfile should be the appropriate file if only input signals from battery voltages and Qualisys sitfpga cRIO 9113 IO CSE lvbitx authored by Senior Engineer Thorgeir Wahl if with forcering sitfpga cRIO 9113 IO CSE Strain lvbitx authored by Senior Engineer Thorgeir Wahl In Options all PWM out Frequency should be set to 50 Hz else strange behavior damage may happen Sometimes when deploying a Conflict Resolution window may pop up this means there is already a previous vi file already deployed and may also be running on the cRIO press Apply will override the old stuff from a previous run If unable to resolve the conflict restart the cRIO by 1 Use Measurement amp Automation program found on desktop and remotly restart cRIO if that fails 2 Disconnect and reconnect all power sources batteries Qualisys Qualisys Oqus The cameras used to register see the IR markers Qualisys Motion Capture Systems is the system that process the data from Oqus Qualisys Track Manager The userinterface to interact with Motion Capture System QTMdriver vi The QT
22. Mdriver is engineered by Senior Engineer Torgeir Wahl to aquire the data from Qualisys It is built in a producer consumer pattern this decouples the capture rate of Qualisys and the mdl solver rate built to handle any number of bodies Qualisys needs to track O index based and passes the signals in the following order 1 Frame number 2 Error signal 78 APPENDIX B DRAFT FOR USER MANUAL x position in millimeter 1 body y position in millimeter 1 body z position in millimeter 1 body yaw in degrees 1 body pitch in degrees 1 body roll in degrees 1 body Residual mean offset between all the IR markers compared to his expected position configuration 1 body 10 x position in millimeter 2 body 11 y position in millimeter 2 body 12 z position in millimeter 2 body 13 yaw in degrees 2 body 14 pitch in degrees 2 body 15 roll in degrees 2 body 16 Residual mean offset between all the IR markers compared to the expected position configuration 2 body etc NAO The driver will send only the lastet newest data queues up all data given from Qualisys and send the newest dataset when data is requested after sending the latest dataset it will purge the queue but hold on the the lastest dataset if capture rate is higher faster than solver rate then no all data recieved from Qualisys will be used if capture rate is lower slower than solver rate then same data will be sent one or several times un
23. NTNU Trondheim Norwegian University of Science and Technology Line Of Sight based maneuvering control design implementation and experimental testing for the model ship C S Enterprise Nam Dinh Tran Marine Technology Submission date June 2014 Supervisor Roger Skjetne IMT Co supervisor ivind Kjerstad IMT Norwegian University of Science and Technology Department of Marine Technology NTNU Trondheim Norwegian University of Science and Technology Department of Marine Technology PROJECT DESCRIPTION SHEET Name of the candidate Dinh Nam Tran Field of study Marine control engineering Thesis title Norwegian Design av regulatoralgoritme for banefolging ved siktlinjemetoden implementering og eksperimentell testing for modellfart yet C S Enterprise I Thesis title English Line Of Sight based maneuvering control design implementation and experimental testing for the model ship C S Enterprise I Background The maneuvering control problem was defined in 2002 providing a novel framework for solving path following problems for a wide variety of dynamical systems By dividing the overall maneuvering problem into a geometric and dynamic task the methodology provides means to construct intelligent control and guidance laws with natural behavior in terms of how a dynamical system solves a path following control objective Within the field of marine technology several applications have been reported in the li
24. PPJoy before LabVIEW aquires them In the current setup the indirect option have limited appliations in terms of control modes However this approach can provde a faster more intuitive and direct control of CSE1 in terms of manual control This is due to the mapping which each Voith Schneider Propellers VSPs s force direction is mapped to an analog stick and the lower shoulder button pair L2 and R2 controls the bow thruster force direction The operator also have the option of choosing where the process programs should run internally on the laptop or externally on the CSE1 The former utilize the mdl file while the latter make use of the rtw files internally and uploads the nidll vxworks rtw files to the CSE1 The nidll rtw and nidll_vxworks_rtw files are derived from the mdl file using Real Time Workshop This thesis focus on the latter option How each signal and controller is treated is handled and structured in a Simulink diagram mdl The communication with CSE1 occurs wireless through HIL Lab Qualisys tracks CSE1 using IR rays and sends the data back to LabVIEW 2 2 CyberShip Enterprise 1 In principle CSE1 consist of a modified hull 1 50 model scale a waterproof box a network adapter D Link a cRIO a bow thruster two batteries two VSPs four servos and several passive IR markers in addition to wires and other electrical components see Skatun 2011 for details 2 2 1 Communication Connection between
25. Systen target file gt Browse Choose nidll tlc gt OK gt Apply gt Build Then Brrowse gt nidll_vxowrks tlc gt OK gt Apply gt Build And you have now _nidll_rtw and _nidll_vxworks_rtw folder in CS Enterprise I models simulink where is the name of the mdl file e g StudentTemplate mdl gives StudentTemplate_rtw and StudentTemplate_nidll_vxworks_rtw Connecting vi file to mdl file Real Time Target 1 In Matlab Simulink create using RealTime Workshop nidll tlc and nidll_vxworks tlc files for the desired mdl file 2 Make sure you are able to communicate with the cRIO on C S Enterprise 1 Check by Command Promt gt ping 192 168 0 77 3 Open SIT Connection Manager vi gt Front Panel gt Tools gt SIT Connection Manager 4 In Execution Host have the Real Time Target selected Target is RT CompactRIO NI cRI09024 CSE1 192 168 0 77 5 Add Targets and Devices on Driver lvproj gt Targets and Devices gt Real Time CompactRIO folder gt NI cRIO9024 CSE1 6 Select Programming Mode gt LabVIEW FPGA Interface 7T Current Model DLL select the dll file created in step 1 inside the folder _nidll_rtw the dll file should have the same name as the mdl file it was created from 8 Change Project Directory to CS EnterpriseI Project folder You might have to redo step 5 again 77 Starting from a clean slate hardware mapping in SIT
26. W here to provide an realtime way to change the matrices values Its output are M D and B M DL B Inertia matrix Linear hydrodynamic dampning matrix Thruster configuration matrix Figure 2 11 C S Enterprise 1 Matrices module in the Simulink diagram 2 6 8 Data logging The point of data log was chosen in Simulink since it is the performances of the control system that are of interest Each barrrier between the point of interest and point of measuring is a potential source of error and delay However this approach requires the operator to manually extract the log file if the mdl file was running on the cRIO through the Measurment amp Automation Explorer The data is stored in mat formate using a oFile block which is generated and stored in the workspace wherever the model is running Execution Host The name of the data file is set in the dialog window of the ToFile block NLPID_data mat LgV2_data mat To File To File Figure 2 12 Two ToFile blocks used in Simulink diagram 26 CHAPTER 2 THE INFRASTRUCTURE Chapter 3 System identification For the system identification arrangement twelve rotation free hook three force ring and three springs were used Two force rings on portside and two springs on starboard side and one force ring in the aft and a spring at the bow In general each measurment series last 30 seconds with 30 seconds in between measurment to allow the surface distubance
27. al representation of the simulations and laboratory runs Adding comments and details on how the laboratory experiments were conducted Editing of the videos taken of the laboratory runs Analysis of the hundreds of data measurments taken Sorting and organizing the attached files Finishing the user manual for CSE1 and include the complementary screenshots Bibliography Breivik M and Fossen T I 2005 Principles of guidance based path following in 2D and 3D in Proceedings of the 44th IEEE Conference on Decision and Control and European Control Conference Seville Spain pp 627 634 Fossen T I 2002 Marine Control Systems Guidance Navigation and Control of Ships Rigs and Underwater Vehicles 1st edn Marine Cybernetics Trondheim Norway Fossen T I 2011 Handbook of Marine Craft Hydrodynamics and Motion Control John Wiley amp Sons Ltd United Kingdom National Instruments n d a Components of a Simulation Simulation Interface Toolkit URL http zone ni com reference en X X help 871504F 01 lvsitconcepts sit c components of a simulation National Instruments n d b Understanding the Driver VI Simulation Interface Toolkit URL http zone ni com reference en X X help 871504F 01 lvsitconcepts sit c understanding the driver vi Norwegian University of Science and Technology n d Marine Cybernetics Laboratory URL http www ntnu no imt lab kybernetikk Skatu
28. and look at a backstepping and a nonlinear control with Line Of Sight based maneuvering control design Another way to describe it is path following through forward speed and heading from two different approaches 1 3 Previous works The maneuvering design approach divides the problem statment into two a Geometric Task and a Dynamic Task The geometric task is to force the output to converge to the desired point or path While the dynamic task is to foce the output to converge to a desired time signal speed and or acceleration It was introduced in Skjetne 2005 and more details can be found there and in Fossen 2011 In Breivik and Fossen 2005 a general framework for path following is presented and in Skjetne et al 2011 the path following is applied as a generic problem with Line Of Sight LOS and the maneuvering approach A step futher is taken in Thorvaldsen 2011 where he explores the possibility of path following in formation using different designs among those the generic maneuvering theory and LOS steering algorithm CSE1 have been used in Sundland 2013 for experiments related to towing of icebergs However several complications made it difficult to produce good results and they are ad dressed and improved upon in this thesis CSE1 was originally developed for demonstra tions and experiments at Marine Cybernetics Laboratory MC Lab and it is documented in Skatun 2011 His framework have been deconstructed and reconstru
29. and z axis points down The orgin is roughly placed along the longitudinal centerline If the basin is empty when calibrating the origin is set around approximately half a meter above the basins bottom If the basin is full when calibrating then it is set a few centimeters abovethe water surface This has to do with the practical aspects of moving placing and retrieving the markers B frame is a body frame for CSE1 The frame moves and rotates with the vessel The origin is in placed along the longitudinal centerline approximately on the longitudinal tipping point determined by balancing CSE1 on a metal pipe and on the waterline The x axis pointing from stern to bow y axis from port to starboard and z axis top to bottom 1 5 Model The general model used for CSE1 to describe the vessel dynamics R d v L 1a MU r Dv 1 1b where r col p v is the position and heading in the O frame p col x y v col u is the velocitiy vector in the B frame Bf is the command force vector in B frame R v is the corresponding rotation matrix M is the inertia matrix and D accounts for the hydrodynamic damping See the section about system identification or Fossen 2002 for more details It is a model suited for ship positioning Fossen 2002 and is suitable for slow speed and calm water applications Another reason is due to limited and inaccurate system identification of the hull The coefficients in the mat
30. andled between servers and softwares The original setup with single frequency was replaced with a multi frequency producer consumer structure The futherst away a vessel is visible for Qualinsys is around 17 meters from the cameras The shortest distance is roughly 4 meters infront of the cameras This result in a lengthwise workspace for a vessel to be in the range of 13 meters This length length can be increased to 27 meters If the carriage start from the beach end and moves toward the wavemaker during the experiment It is not possible to increase it futher due to the length of the basin the wavemaker beach in outlet and carriage is occupying In the system identification an addtitonal hydrodynamic damping coefficent for slow speed have been determined for the hull N 0 18140 In addtion higher order coefficent have also been estimated from the towing measurements A pseudo library for lookup table thruster mapping have been created The bow thruster have been mapped for power limit input 0 15 0 3 0 4 0 5 and the voith schneider propellers for speed input 0 3 0 4 Measurments for voith schneider propellers for speed 0 2 have also been conducted but no lookup table have been created for it The starboard voith schneider propeller rotates slow than the port voith schneider pro peller It is also noted that the servos are significantly coupled and the force output drops at the periphery value 1 Of the servos servo 4 ha
31. arallel Port Joystick software designed to add virtual joysticks under windows operating systems 4 8 11 13 Qualisys Motion capture system 3 4 8 11 13 14 24 25 39 Real Time Workshop Generates C C code from Simulink diagrams mdl files re branded as Simulink Coder in later versions 4 8 16 40 servo device used to provide control of a desired operation through the use of feedback 8 9 Simulink Graphical programming language developed by MathWorks 4 8 13 15 17 20 23 25 39 40 Simulink Coder Rebranded name of Real Time Workshop 40 TeXworks Program used for writing in LaTeX 4 vxworks Real time operating system developed as proprietary software by Wind River Systems designed for use in embedded systems 13 Acronyms BT Bow Thruster 21 22 27 28 44 cRIO Compact Realtime Input and Output 4 8 9 14 24 25 CSE1 CyberShip Enterprise 1 1 5 7 14 25 27 31 39 46 71 DOF Degrees Of Freedom 4 18 20 21 43 DP Dynamic Positioning 12 22 FPGA Field Programmable Gate Array 4 10 14 GNC Guidance Navigation and Control 4 24 GUI Graphical User Interface 4 8 10 12 IR InfraRed 3 8 10 LgV Control Lyapunov Function 12 LgV2 LgV backstepping 2 5 22 LOS Line Of Sight 2 18 20 21 43 MC Lab Marine Cybernetics Laboratory 2 5 7 8 10 11 39 40 MSS Marine Systems Simulator 4 24 NLPID Nonlinear PID 5 22 PID Proportional Integral Derivative 1 5 22 41 PS PlayStation 4
32. ariable with a unique variable name and use a From block in the Control Switch module to retrieve For automated controls reference values tuning parameters 7 and v Control 0 is used for the direct thruster control through a PS controller It need the x and y coordinate of each analog stick R2 and L2 signal and BT power limit and VSP speed setpoint to create Te For a DP PID controller The desired position and orientation ng vessel dynamic matrices M and D tuning matrices Kp Kr and Kp and 7 and v are needed as inputs For the LgV2 and NLPID design from Skjetne 2014 all the outputs from Guidance a control reset variable called ctrlReset intial values so and qo vessel dynamic matrices M and D tuning parameters and matrices Ly A Kp KK and Kp and 7 and v are needed as inputs Their outputs are 7 virtual point q and path parameter s Control Switch The Control Switch module is a straight forward subsystem with a Switch block It takes in the T s from the controller modules and a variable named controlModeS elector to decide which controller to use The Switch block is zero based since the ControlMode radio button in LabVIEW is zero based Thruster allocation The Thruster allocation module converts into Te in two steps First faci Bw 2 4 where face col fi fs fa fs is the force actuator vector is the force vector in B fram
33. cRIO and LabVIEW is of the utmost importance without it remote control of CSE1 is impossible An option to ensure a near 100 connection is to have a ethernet cable connecting the two together However this would severely cripple the mobility practiclly mooring it The other option is wireless connection All communication with CSE1 takes place wirelessly through the D Link In the early stagest loss of connection with CSE1 was a common occurrence This had to do with where it was relative to where the HIL Lab router was and where on the hull the D Link was placed Originally the standard antenna was used for the D Link and placed within the watertight box bending the antenna almost parallel to the cRIO The combination of these naturally deteriorated strength of the wireless connection The loss of connection was lessen with the aid of Senior Engineer Torgeir Wahl The first action adding an additional wireless Asus placed more centrally and closer to the space CSE1 operated However this solution was short lived since the Asus died after periode of use The cause of this is unknown The next action gave a more permanent fix to the problem The antenna was moved outside the waterproof box while the D Link remained 2 3 MARINE CYBERNETICS LABORATORY 9 inside This was achieved by attaching the D Link to the inside of the waterproof box s lid with velcro tape then drill a hole connect the antenna and sealing it with caulk Involu
34. check wiring Check battery voltages Main battery should be 10 Volt or more maximum around 13 Volt regular 11 to 12 Volt low 10 Volt Servo battery should be in 5 Volt or more max around 6 4 Volt regular around 6 Volt Note Black wire should always be the last to be connected and Main Battery first Using PS3 controller BtSix exe need to be running and can be found in the folder CS Enterprise I PS3Control BtSix 1 5c BtSix must be active and connected to the PS3 controller prior to Deployment of the of the vi file Tilting the PS3 controller should give some respons in the BtSix window Make sure the Joystick info have under Device Name PPJoy Virtual else the signal might not get through Scroll to channels untill you get it if it is no the right one The boolean for the buttons and sticks should light up when PS3 controller is used Avoiding clutter in CS Enterprise I folder Set Project Directory in SIT Connection Manager as it will create five files sithwconfig Driver aliases Driver lvproj _Driver vi and _I0 11b for each mdl file connected with SIT Connection Manager Creating nidll tlc and nidll vxworks tlc files 76 APPENDIX B DRAFT FOR USER MANUAL In Matlab gt Simulink gt mdl Set Current Directory to CS Enterprise I models simulink Simulation gt Configuration Parameters Ctrl E gt Real Time Workshop gt Target selection gt
35. cted in prepera tion for this thesis see extract of Tran 2013 Instead of quoting half of Sk tun s thesis it should be read before reading this thesis as it is the foundation buildt upon This thesis is a continuation where it will further develop and elaborate on the mechanics of CSE1 s framework 1 4 Preliminaries 1 4 1 Notations The notations corresponds with Skjetne s and Thorvaldsen s work Time derivatives of z t are denoted as 3 x9 c while partial differentiation 1 0 t o z 0 t m and a z 0 t a The Euclidean vector norm z a7 2 and stacking vectors into one is denoted as col z y z x7 yT zT T The subscript d stand for desired 1 5 MODEL 3 1 4 2 Reference frames O frame is a inertial reference frame within MC Lab similar to a North East Down frame It is used by Qualisys to determine the position and orientation of the bodies observed The origin and orientation of the frame is set when calibrating Qualisys This is done using two sets of InfraRed IR markers where in each set the markers have fixed placement relative to eachother One set of markers is placed on a fixed position specifying the origin and orientation of the frame while the other is moved around to calibrate the the the workspace Normally the x y and z axis are parallel the basins walls where x axis points toward the wavemaker y axis points away from the walkway
36. e and is the pseudoinverse of the thruster configuration matrix B Then f is mapped to thruster inputs u through lookup tables before merging with the BT power limit and VSP speed setpoint to create the thruster commands set T e If other power limit or speed setpoint is desired by the user then the lookup tables needs to be replaced with the appropriate ones This part is hardcoded into the control architecture For details on the lookup tables see system identification section 2 6 SIMULINK 23 2 6 5 Plant The Plant module s main function is to make use of and Te and directly or indirectly produce 7 and v It is divided into two subsystems Real Target and Simulator Real Target Simulator Figure 2 9 Plant module in the Simulink diagram Plant module input controlModeSelector Parameter for switching between controllers enableCSE1 Parameter to enable the thruster subsystems LS Enable Parameter to enable the Linear simulator y Force vector in B frame T Thruster commands set TC n Direct thuster command set from control M Inertia matrix D Hydrodynamic damping matrix LS Reset Parameter for reseting linear simulator No Initial values of nrs Vo m Initial values of vzs Plant module output NLS Position and heading form linear simulator VLS Velocities from linear simulator Real Target The Real Target use
37. e a CR E LAL Notations 2x22 ns Oe GER Ew Ree EE Da ai 1 42 Reference frames Model ak uu od aseo Roo E Behe PR BAe ee eed hee ow X DOIUWBEB ORs dedodeX mms d RR X Edo SCODE Ca ks b Eon ee bee EEG A nome equus E dede gos heed wingidn AENEID Infrastructure uu ouo w Mow E Ron kg GRECE des ak My Rm mo uod Gybership Enterprise L cf ce eae baa kG Ree RR s 221 ComnmrmiunicdblONH 220 weskuac4 4e wee wovow EGSoR OY o XX ow 222 VISBY one RAR EMER PRA Rs d 223 DBRUSEBES luo DA Se ek ikea WEW oa eG Marine Cybernetics D DH PRI LITT LADVIBAW s waw how eo Gobo ee ee DS ME P0 e ae E 25 2 From Panel 2 2 vo RR do Vo e Dee Md 2 5 3 Bleck Diagram awoo osiowy ee ea ee eS 2 5 4 Simulation interface toolkit 2 5 5 Qualisys Track Manager Drivers siu WG Mow B R ets d B R he EA 206 1 Input trom LabVIBW onc eee aa eG xxx e ek ah hai 2 0 2 Output te LabVIEW eb uem o RE RR 263 Gudane tu Ee RRR Eh RRA Eee wea Sb Ic eek cee ee OS Re 4 xe vii viii CONTENTS WE ro MMC 23 ZM NAWEAN W oe s ma s ede x de Boe a Be Arte Ra e ee a 24 2 6 7 C S En
38. eller rotates slow than the port voith schneider pro peller The effect is notiable in the meaurment However it does not come into effect unless 3 2 THRUSTER MAPPING 29 Force acting in surge direction on Cybership Enterprise from towing Measured dampning force Fitted curve X u lulu Ku Fitted curve X u DP 0 19227 2 81407 0 37454 0 59739 Mean drag force N 0 8 0 6 0 4 0 2 0 0 2 0 4 0 6 Surge speed m s Figure 3 3 Measurments of dampning coefficient of CSE1 the maximum force are required and it becomes less significant as higher rotation speed are used It is also noted that the servos are significantly coupled and the force output drops at the periphery value 1 It can be seen from the thruster measurements that the lookup tables for the voith schneider propellers does not truly cancel out the circular motion of the servos Some of it can be caused by the rotation when measureing or due to placement of thuster However this can not fully explain the notable force meaured in the other direction Of the servos servo 4 have the strongest coupling in surge sway During the meaurment of the thrusters it is noted that the thuster commands have an error of margin around 0 03 Meaning the 0 21 0 22 and 0 23 can generate the same force See appendix for plots 30 CHAPTER 3 SYSTEM IDENTIFICATION Force acting in sway direction on Cybership Enterprise fram towing Mea
39. ery seriously by the university and will have consequences NTNU can use the results freely in research and teaching by proper referencing unless otherwise agreed upon The thesis shall be submitted with two printed and electronic copies to 1 the main supervisor and 2 the external examiner each copy signed by the candidate The final revised version of this thesis description must be included The report must appear in a bound volume or a binder according to the NTNU standard template Computer code and a PDF version of the report shall be included electronically Start date August 2013 Due date As specified by the administration Supervisor Roger Skjetne Co advisor s ivind Kjerstad PhD candidate Trondheim Roger Skjetne Supervisor Summary This report documents the progress methods and engineering in building and testing the framework for CyberShip Enterprise 1 The main focus have been to modularize standard ize and improve the infrastructure and the operative system with respect to performance With a secondary objective to create a user manual to operate it The work is done for a surface vessel in 3 degrees of freedom surge sway and yaw and in calm waters and slow speed The reliability when conducting laboratory experiments with CyberShip Enterprise 1 have greatly improved This was achieved through repositioning and replacing the antenna for wireless communication and a restructuring of how signals were h
40. es it on to the HIL Lab network 2 5 LabVIEW As mentioned previously LabVIEW routes and displays the data The Front Panel pro vides the GUI and the Block Diagram and SIT for signal routing The programs have been deconstructed and reconstructed several times The same vi file can be used for several mdl file e g Template HMI vi are used for both TemplateNIPID files and TemplateLgV2 files SIT will generate separate project files based on the mdl file name The drivers and FPGA files were orginally created by Senior Engineer Torgeir Wahl some of those have later been modified by the author The programs are intended for single vessel body with the possibility to expand for multiple bodies by modifications The focus will be on single body 2 5 1 vi files Four vi files have been created in connection with this thesis PS3 HMLvi Thruster HMLvi StudentHMI vi and Template HMI vi Each of them shows different stages 2 5 LABVIEW 11 Figure 2 2 Qualisys Oqus area covered seen from above The Oqus are placed on right edge white cones indicate visible area per cameera The walkway is at the bottom edge and somewhere on the left edge is the wavemaker Red arrow indicate x axis and teal arrow indicate y axis Each square in the grid is 0 5 x 0 5 meters of the development PS3 HMI vi was the first one created directly derived from Skatun s work It contains and require the bare
41. f the cameras This result in a lengthwise workspace for a vessel to be in the range of 13 meters This length length can be increased to 27 meters If the carriage start from the beach end and moves toward the wavemaker during the experiment It is not possible to increase it futher due to the length of the basin the wavemaker beach in outlet and carriage is occupying In the system identification an addtitonal hydrodynamic damping coefficent for slow speed have been determined for the hull N 0 18140 In addtion higher order coefficent have also been estimated from the towing measurements A pseudo library for lookup table thruster mapping have been created The bow thruster have been mapped for power limit input 0 15 0 3 0 4 0 5 and the voith schneider propellers for speed input 0 3 0 4 Measurments for voith schneider propellers for speed 0 2 have also been conducted but no lookup table have been created for it The starboard voith schneider propeller rotates slow than the port voith schneider pro peller The effect is notiable in the meaurment However it does not come into effect unless the maximum force are required and it becomes less significant as higher rotation speed are used It is also noted that the servos are significantly coupled and the force output drops at the periphery value 1 It can be seen from the thruster measurements that the lookup tables for the voith schneider propellers does not truly cancel out
42. he mdl file had a higher frequency than Qualisys the Base Rate Loop vi would crash ending the run This is caused by the driver not having any data to pass on and nothing is sent to the mdl The orignal driver also contained an error where the size of the output array was smaller then the actual size This caused the shuffling of the data set Even if both Qualisys and mdl were both set to the same frequency it would have only been a matter of time before mdl in a time step began ahead of Qualisys This is due to that each of them have their own internal clock that is not synchronized When the difference between the two becomes too large Base Rate Loop vi crashes Most runs could not last longer than a few minutes and this was the main problem As each crash meant a total reboot of LabVIEW combined with the lost of connection most of the time was spent on establishing and deploying the software To fix this problem a Producer Consumer desgin pattern was implemented The Pro ducer QTMTask vi and Consumer QTMdriver vi replaced the orignal QTMdriver vi QTMTask vi follows Qualisys freqency and QTMdriver_ vi follows real time target fre quency mdl file This setup makes them frequency independent of each other QTMdriver vi aquires the data using other vi files and add it to a shared memory block data queue it have with Q TMdriver vi QTMdriver_ vi retrives the data set from the data queue wipes the queue clea
43. in Simulink using SIT Connection Manager in the vi file TemplateLV_HMI vi See previous note on connecting vi to mdl The SIT Connection Manager should have created several files in the chosen designated Project Directory with filenames begining with the mdlfilename e g TemplateSL mdl will produce TemplateSL sithwconfig TemplateSL Driver aliases TemplateSL_Driver lvproj TemplateSL_Driver vi TemplateSL_I0 11b Open the input output library e g TemplateSL_I0 11b Open baserate loop vi e g TemplateSL_Base Rate Loop vi From Front Panel go to Blockdiagram Ctrl E or Windows gt Show Block Diagram Add the QTM driver inside Init Code Read Code and Close Code Flat Sequence Structure Gray frames below the name tags NB do not add the QTM driver to the Write Code Flat Sequence Structure Add QTM driver by inside the Block Diagram 1 right click to get function curtain 2 Click on Select a VI 3 Go to QTMdrivers folder and select appropriate driver e g QTMdriver OneBody vi Tip add one as decribed above and select the added block hold Ctrl and drag it to copy it Connect the vi s together 80 APPENDIX B DRAFT FOR USER MANUAL Tip 1 Have Context Help window open Tip 2 right click on the vi block and uncheck View as Icon Tip 3 Ctrl U or Edit gt Clean Up Diagram will automatically organize the wire and blocks for you if you dont like it press C
44. interacts with the rest of the programs when using CSE1 The panel is directly connected to the Block Diagram and vertcally divided into two parts Show Virtual ship vom e w mne gt mas 30 Visualization Regulation Command Scope Information es cwe fo control Enable Linear Simulato QS Frame Qs Error gt o ps3 Reference based control Setpoint Path xCSEL m yCSEL m psiCSEL deg a ZEE 357224 217611 84 6901 v Ji gio xAux m yAux m psiAux deg 4 4 07697 12501 23 654 ul xdlm psidIdeg a1575 136026 asa Linear Elipse Limiter ulm s vim rldeg s yom 0 02959 o 07218 2641 95 so Main Battery 10 64 Servo Battery 5 44235 Bowthruster Battery 16 17365 2032344 PENES 59 n a L i e p Figure 2 3 Overview of LabVIEW Front Panel The left part contains the dials switches indicators and controls for input and selections To switch between the different control mode use the radio button column located on top in the middel It is mapped to a constant block in the mdl file sending an integer from 0 and up That value is then used in a switch block to determine which input to use Similarly for which input to pass to the controller the radio button column above the Enable Linear Simulator boolean button The tabs
45. l send the signal which the Simulink diagram will update with before executing The values set in the Gain blocks will be multiplied with the signal before sending it to LabVIEW if mapped Therefore most often those values are set to one The latest versions utilize GoTo and From blocks with global variables to pass values between subsustems The previous versions used wires to send signals between the subsystems This created a lot of clutter and unnecessary work when adding removing or just moving blocks due to the path the wires would be automatically placed E gt D Constant Gain From Goto Figure 2 5 Blocks used for signal routing and mapping in Simulink diagram This enables most blocks to be placed anywhere in the Simulink diagram However for structure and logical flow when looking at the it the blocks are placed as if they were using wires and parts placed in subsystems where it is logical In general the diagram can be read from top to bottom and from left to right The top level contains four blocks where three of them are subsystems The SignalProbe block is the port for communication with the SIT server All values mapped from Lab VIEW are gathered in the subsystem Input from LabVIEW Every signal mapped to LabVIEW are located in the Output to LabVIEW subsystem Each inputs from SIT are in the subsystem Input from SIT found under Main Subsystems Navigation Outputs to SIT
46. lays it sees all markes but it is unable to calculate the vessels position and orientation The cause of this needs to be further investigated Spare part servo arms should be purchased and padding for the hull to dampen the impact to the basin walls Several crack have been observed on the hull Simulation Interface Toolkit have been discontinued modifying it for Modular Interface Toolkit can be an option The labeling and choice of variable name can be improved upon Preface This thesis concludes my M Sc studies not what I expected but more than what I could hope for The purpose of this report was to document the work carried out in the Marine Cybernetics Laboratory at the Norwegian University of Science and Technology since Fall 2013 The original version of this report and content was lost durring transit as a result of several unfortunate decisions What is presented here is a shell of the original recreated during the last two weekends before the deadline In hinsight hubris was the reason as they say pride come before the fall The work began with reviewing the previous work done on CyberShip Enterprise 1 be fore restructuring and improve it This lead to a more modularized and generic control architecture suitable for many types of control designs In relation to this work countless days have been spent in the Marine Cybernetics Laboratory conducting measurements of various parameters related to controlling CyberShip Enterprise 1
47. literature study on applications of CSE1 in projects and papers on the maneuvering control design method and on LOS based control designs Write a list with definitions and descriptions of relevant terms and concepts 4 Let p x y be the position of a point mass with a double integrator dynamics i e u Leta desired path be a straight line to be traversed with unit speed Implement and simulate a maneuvering control law for this system and explain its behavior in terms of how and why the filtered unfiltered gradient update laws work 5 Propose how to implement a LOS based maneuvering control mode within the control system architecture for CSE1 with functionalities for setting the path specifying the speed along the path and to get feedback on the actual motion of the ship relative to the desired path in the lab 6 Design a guidance system and a LOS based maneuvering control law based on full actuation of CSEI The path given by the guidance system should utilize the space in MC Lab as much as possible without needing to terminate early the operation due to space constraints To consider if one can move the carriage during experiments to get more space available Present simulation results for the system NTNU Faculty of Engineering Science and Technology Norwegian University of Science and Technology Department of Marine Technology 7 Implement and test the LOS based maneuvering control system for CSE1 in MC Lab and present the
48. m blocks obtaining the signals from the global variables and divides most of them into scalars for mapping The blocks are the counterpart of the Indicators in LabVIEW 2 6 3 Guidance The Guidance module s main function is to generate all the reference or desired variables the controllers needs It contains e a Path Heading Speed assigment and Line Of Sight module 2 6 SIMULINK 17 Path Heading Speed assigment Line Of Sight Figure 2 7 Guidance module in the Simulink diagram Guidance module input pathSelector m s m m Switch workaround parameter for path Path parameter of desired path p x coordinate of origin of ellipse path in O frame y coordinate of origin of ellipse path in O frame Radius of ellipse path in x direction in Q frame Radius of ellipse path in y direction in O frame Scaling parameter of path parameter s x coordinate of linear path in O frame when s is zero y coordinate of linear path in O frame when s is zero x coordinate defining heading of linear path in O frame y coordinate defining heading of linear path in O frame Desired surge speed in 5 frame Virtual point mass coordiantes of vessel in Q frame Lookahead distance Tuning parameter for s gradient algorithm 18 CHAPTER 2 THE INFRASTRUCTURE Guidance module output x col m m rad 3DOF reference vector in O f
49. n H N 2011 Development of a DP system for CS Enterprise I with voith schneider thrusters Master s thesis Norwegian University of Science and Technology Skjetne R 2005 The Maneuvering Problem PhD thesis Norwegian University of Sci ence and Technology Skjetne R 2014 Report Maneuvering los control design The fully actuated case in preparation Rev E Skjetne R Jorgensen U and Teel A R 2011 Line of sight path following along reg ularly parametrized curves solved as a generic maneuvering problem in Proceedings of the 50th IEEE Conference on Decision and Control and European Control Conference Orlando Florida USA pp 2467 2474 SNAME 1950 Nomenclature for treating the motion of a submerged body through a fluid in Technical and Research Bulletin No 1 5 37 Sundland M N 2013 Guidance and control of iceberg towing operation in open wa ter with experimental testing Master s thesis Norwegian University of Science and Technology Texas A amp M University n d HOW TO CONNECT SIMULINK TO LABVIEW IN ORDER TO COLLECT SYSTEM DATA URL Attp parlos tamu edu MEEN364 Simulink2Lab VIE W pdf Thorvaldsen C F L 2011 Formation control of marine vessels Master s thesis Nor wegian University of Science and Technology Tran N D 2013 Development of a modularized control architecture for cs enterprise i for path following based on los and maneuvering theory Technical repo
50. n passes the data set on to the SIT server and stores the data in a memory block If there is no new data set most of the data set in the memory block is passed to the SIT server instead analog to a zero order hold This setup greatly improves the reliability and robustness of the system For more information National Instruments n d b and National Instruments n d a 2 6 Simulink The control architecture is defined in the Simulink diagram As previously mentioned the mapping between Simulink and LabVIEW is handled by SIT The blocks utilized for the actual routing are Constant blocks for signals from LabVIEW Controls and Gain blocks for signals to LabVIEW Indicators The names given to the Controls in LabVIEW are set to be similar if not excatly the same as their counterpart in Simulink Constant blocks An example The Control that determines which controller to use is called Mode Control in LabVIEW It s counterpart meaning the Constant block it shall be mapped to in the Simulink diagram is given the name Mode Control Selector This simplifies the process when the mapping is done 2 6 SIMULINK 15 in SIT Connection Manager making it easier to know which Control or Indicator to mapped to which block The values set on the Constant blocks are often set to the default value preferred However those values does not matter when running the it via LabVIEW since the vi wiil
51. n full scale and out in the field can be both a costly and time consuming investment of resources A more reaonable practical and effective way is to perform laboratory experiments since it is down scaled both in terms of size and cost It can provide a proof of concept beyond just simulations and a stepping stone towards a full scale experiment This thesis aims to advance one of these great ideas Line Of Sight based maneuvering control design and hopefully provide an easy framework for others to work on 1 2 Background The model based ship control began with the introduction of the gyrocompass in 1908 and was further developed as other positioning systems became available Another way to look at is the dawn of autopilots The purpose is to carry out operations or maneuvers without constant human intervention It can be applied surface and underwater vehicles Examples of this can be station keeping weather optimal positioning and tracking There are various ways to attain these objective it can be through a simple Proportional Integral Derivative 2 CHAPTER 1 INTRODUCTION PID linear or nonlinear with feedback feedforward neither both or just one of them Linear Quadratic Optimial Backstepping Sliding Mode Control and several others A path can be parameterized discrete continuous or a hybrid of those for details on these topics see Skjetne 2005 Fossen 2002 and Fossen 2011 This thesis will use a continuous path
52. ntarily disconnection frequency lessen but still a nuisance The last improvement was to to replace the standard antenna that followed the D Link with an antenna roughly three times the original s length Spontaneous disconnection became more or less extinct However it is worth noting that the probability of establishing connection is mostly deter mined by the sequence the various wires are connected to the batteries From experience without any scientific documentation connecting the red wire before the black requires fewer attemps of establishing connection than vice versa 2 2 2 Visibility CSE1 s postion and orientation are aquired by Qualisys using the IR markers placed on board Therefore size and placement of the markers have an impact on CSE1 s visibility Orignally a fixed IR marker stand was used and placed on top of the waterproof box along the longitudinal centerline Qualisys had at times and certain orientation relative to the cameras lost sight of CSE1 due to the relativ positions of the markers to each other and their proximity to the latest antenna Sometimes the markes merge together other times they overshadows one another in addition to the antenna overshadowing or dividing the markers To increase the visibility elevated passive IR markers were added two on the longitudinal centerline bow and stern three on the aft end two port one starboard The first setup had the fixed stand and the five markers a total
53. ntroduce a delay The controllers are designed with three points that chases each other the path point ny the virtual point x and the vessels point 7 1 stays on the path and moves to minimize the distance between it and x x tries first to minimize the distance between it and glseta before na While 7 will only converge to x If the filltered update law is disabled controller becomes a tracking case Attached electronically are the data from the runs burt due to timeconstraint those are not presented here 4 1 Simulation runs In the ideal simulated word both of them were equal in terms of maneuvering both for linear paths and for ellipse path Depending on the tuning their inital transient behavior can be erractic and unnatural 4 2 Laboratory runs Originally only the LgV backstepping was tested in the laboratory It converged and per formed well The only downside was it would constantly overshoot the heading resulting in a constant oscillation while moving alone the path The reason was due to the noisy velocity estimation 33 34 CHAPTER 4 LINE OF SIGHT EXPERIMENTS Other opportunity presented itself to run more experiments in the laboratory However unknown at that time and discoved to late one of the servo arms broke due to fatigue from the constant oscillation of the previous runs corruping the mapping and observer estimates In the laboratory their performance depended heavily on how they were imple mented If bo
54. ong desired path in O frame vector col z y4 17 18 22 44 q Virtual point mass coordiantes of vessel in O frame 17 18 20 22 Yaw angular velocity about z axis in B frame radian per second rad s 3 43 R v 3 x 3 rotation matrix between Q frame and B frame 3 r Radius of ellipse path in x direction in O frame m 17 19 ry Radius of ellipse path in y direction in O frame m 17 19 s Path parameter of desired path py 17 22 43 Thruster commands set col u BT power limit VSP speed 20 22 24 u Surge linear velocity in x direction in B frame meter per second m s 3 43 uq Desired surge speed in B frame m s 17 19 44 u Thruster input signals col ui u us 4 22 44 v Sway linear velocity in y direction in B frame meter per second m s 3 43 vs Speed assigment corresponding to wa m s 19 X Force in z direction N 43 x Position in O frame in meters m 3 44 zo Origin position of ellipse path in O frame in meters m 17 19 Start position of linear path in O frame in meters m 17 44 Direction position relative to x of linear path in O frame in meters m 17 za Desired position in O frame in meters m 44 Y Force in y direction N 43 y Position in O frame in meters m 3 44 yo Origin position of ellipse path in O frame in meters m 17 19 y Start position of linear path in O frame in meters m 17 44 y2 Direction position relative
55. ort to right presized input array port 7T From Read Code Flat Sequence Structure vi block port to Close Code Flat Sequence Structure vi block port Connection ID out to Connection ID Turquoise vire QTMqueue Out to QTMqueue Orange with turquoise shell wire VI Refnumb QTMTask Out to VI Refnumb QTMTask light turquoise wire 8 Save and exit
56. rame col q ios xa Partial differentiation of x x Partial differentiation of x n rad m Partial differentiation of 2 m rad m Partial differentiation of Vros s rad m Partial differentiation of Vios i s rad Partial differentiation of Vros Ak rad Partial differentiation of Vros Js Dynamic LOS assigment for q a Partial differentiation of f B Partial differentiation of f Partial differentiation of f fs Dynamic LOS assigment for 1 Partial differentiation of f gt Partial differentiation of fs E Partial differentiation of fs Path The Path module s main function is to generate desired position pq Continuous pa rameterization of the paths are implemented in this module It have three modules one for linear path one for ellipse path and a workaround switch Both paths are created simultaneously using the same path parameter s However s will only be dependent on the chosen path uperscript The linear path is created with xz 2 21 y y2 y1 s y y2 y where x and y are coordinates of linear path in Q frame when s is zero and and y2 defines the heading of linear path in Q frame The higher order partial differentiations are set to 0 The ellipse path is created with 2 6 SIMULINK 19 rz cos ks zo 2 2a W ryksin ks 2 2b g pk cos ks 2 2c g rk sin ks 2 2d y ry sin ks yo 2 2e y rykcos ks 2 2 y ryk sin ks
57. rement and generating graphs Simulink is used for creating and modifying the control architecture Real Time Workshop is for convering the Simulink diagram for real time experiments Various blocks from MSS GNC Toolbox are used in the diagram Simulink version 8 2 have been used for creating some of the pictures found in this report QTM is needed for measuring position and orientation in the MC Lab BTSix and PPJoy are used for signal processing of the PlayStation PS controller MCG Reg 4 0 was used for logging force ring data and TeX works for creating this report For a simple introduction in how LabVIEW and Simulink are connected see Texas A amp M University n d 1 7 Scope As previously mentioned the focus is on surface vessel in calm water and slow speed This means 3Degrees Of Freedom DOF surge sway and yaw and linear dynamics This thesis primarily centers around the infrastucture of CSE1 going into detail about the various resources available Two different approaches are implemeted for path following within 1 8 STRUCTURE 5 the control architecture They are the LgV backstepping 2 LgV2 and the Nonlinear PID NLPID design from Skjetne 2014 The behavior of the controls will be compared to each other However details of the control designs wont be covered here just refered to by the equation numbers Description and comments of the infratructure components as well as the reason behind them will be documented as far as
58. rices follows the notion of The Society of Navel Architects and Marine Engineers SN AME 1950 and are similar to those found in Skjetne 2005 cos w sin v 0 R sin d cos 0 1 2 0 0 1 The inertia matrix Xu 0 0 0 m Y mz Y Y 1 3 4 CHAPTER 1 INTRODUCTION The hydrodynamic damping matrix ow 0 D 0 Y YX 1 4 0 Ns N The thruster configuration matrix is the same as in Sk tun 2011 1 B 0 1 0 1 1 1 5 ly li ly2 122 l3 However the coeffficent values have slightly change face col fi fa fa fa fs is the force actuator vector that needs to be mapped to thruster inputs u See the system identification section for details 1 6 Software The softawares used in this thesis are LabVIEW 2010 service pack 1 with Field Programmable Gate Array FPGA Real Time and Simulation Interface Toolkit SIT module MAT LAB 2009b with Simulink Real Time Workshop and Marine Systems Simulator MSS Guidance Navigation and Control GNC Toolbox Qualisys Track Manager QTM BT Six PPJoy MCG Reg 4 0 and TeX works LabVIEW was used to create and is the Graphical User Interface GUI to operate CSEI The FPGA module is for signal handling within Compact Realtime Input and Output cRIO The Real Time module is for running the programs in realtime The SIT module is for connecting LabVIEW together with Simulink and Qualisys MATLAB was used for post processing force ring measu
59. rriage It s motion is limited to the rails along the basins length that ends approximately at the edge of the beach All of these machinery and facility takes up space The in and outlet with the beach occupy roughly four meters of the total basin length around the furtherst position the towing carriage can be placed The towing carrage itself fills up four meters of length Lastly the wavemaker takes up about one meter Using 39 meters as the total length of the basin a surface vessel should have approximately 30 meters of basin length to maneuver in 2 4 Qualisys Qualisys is the real time positioning system available in the MC Lab It uses three IR cameras Oqus to capture the motions They are mounted on the towing carriage front one in the middle and one on each side slightly tilted down Oqus sends out IR rays which is reflected by passive IR markers If the reflected rays are registred by two or more cameras and they are able to clearly identify three or more markes then the position and orientation is captured The visible field of each Oqus is cone shaped Due to height placement relative to water surface the closes visible areas are roughly three meters in front of the cameras Taking the tilted orientation of the cameras relative to the water surface size and proximity of the markers the effective range is around 19 meters QTM is the software that process the data from Oqus and send it to a Qualisys server who in turn pass
60. rt Glossary dll Dynamic Link Library 13 mdl MoDeL file extension Models created with Simulink 8 10 14 16 25 40 vi Virtual Instrument file extension Basic building block for programs written in Lab VIEW 10 15 39 B frame CSE1 s body frame 3 17 21 23 43 44 Q frame Qualisys inertial reference frame in MC Lab 3 17 22 43 44 Base Rate Loop vi Subprogram created by SIT for input output of data 13 14 Block Diagram Part of a vi file containing Control Terminals Wires Structures and various nodes The main window used when constructing a LabVIEW program 8 10 12 14 Bluetooth Wireless technology for exchanging data over short distances 8 11 39 BTSix BlueToothSix software that enables use of a Bluetooth PS controller on a com puter 4 8 11 13 Dongle Small piece of hardware that attaches to electronic devices enabling additional functions 8 11 Front Panel Part of a vi file containing Controls and Indicators The main window used when running a LabVIEW program 8 10 11 13 HIL Lab Hardware In Loop Laboratory local ethernet in the MC Lab 8 10 Program created by SIT for input output of data 14 LabVIEW Graphical programming language developed by National Instruments 4 8 10 16 22 25 39 MATLAB Matrix laboratory high leve technical computing language developed by Math Works 4 8 MCG Reg Program used for logging data in the MC Lab 4 39 40 Glossary PPJoy P
61. s LabVIEW yu Hast computer Figure 2 1 General overview of communication CSE1 was buildt with the intension for use the MC Lab According to Norwegian University of Science and Technology n d the MC Lab was a storage tank for ship models made of paraffin wax and operated by the Department of Marine Technology It contains a basin with wave making towing and real time position measuring capability In addition it have equipments for measuring forces and wave heights At the time of use the computers in MC Lab were running on Windows XP limiting the use of software versions toLabVIEW 2010 Service Pack 1 and MATLAB 2009b As a consequence the programs developed in this thesis were created for compatability and does not take advantage of certain simpler and more advance functions available in newer versions The support for Windows XP expired in April 2014 meaning the operating systems needs to be upgraded to Windows 7 lifting the software version restriction 7 8 CHAPTER 2 THE INFRASTRUCTURE CSE1 is operated from a laptop using a LabVIEW Front Panel as the GUI and the Block Diagram and SIT for signal routing The operator have the option of interacting directly with CSE1 through the Front Panel or indirectly through a PS controller With the indirect apporach the PS controller communicates with LabVIEW via Bluetooth The signals are received by aDongle and processed first by BTSix then by
62. s the output from the Control module and convert them into signals that each thruster is able to follow It needs three input signals T controlMode Selector and an enabling variable enableCSE1 to enable the thruster subsystems The creation and tuning of those subsystems are documented in Skatun 2011 According to 24 CHAPTER 2 THE INFRASTRUCTURE Skatun the 2D lookup tables should counteract the circular motion of each servo and create a linear movement of the VSP control sticks The only thing different from the original is the modularize structure There is a workaround for the T to account for controllers that are direct thruster controls TC n if they are implemented in the Control module However manual adjustment and check is needed to make sure the workaround corresponds to the right controller This subsytem does not directly produce 7 and v since it just sends command signals to a cRIO who in turn routes the signal where they need to go The and v are calculated in the Naviagtion module Simulator The Simulator does not make use of the thrusters instead it runs on a linear vessel dynamics model presented in the introduction equation 1 1 It needs 7 M D an enable parameter LS Enable a reset parameter LS Reset intial position and heading no intial velocities vg The outputs are yrs and vrs 2 6 6 Navigation The Navigation module s main function is to calculate
63. sured dampning force Fitted curve YoY Myy MY Fitted curve Y v DP gt w o A 5 c Y 7 88518 hd v ca c Y wv 55 60856 a Y 3 50625 DP 0 8 0 6 0 4 0 2 0 2 0 4 0 6 Sway speed m s Figure 3 4 Measurments of dampning coefficient of CSE1 3 2 THRUSTER MAPPING 31 Moment N acting in in yaw mode on Cybership Enterprise from towing Measured dampning force N 0 10592 Fitted curve N v 1 25399 Fitted curve N v DP N 0 18140 oP Mean drag moment N 0 8 0 6 0 4 0 2 0 2 0 4 0 6 Sway speed m s Figure 3 5 Measurments of dampning coefficient of CSEI 32 CHAPTER 3 SYSTEM IDENTIFICATION Chapter 4 Line of Sight Experiments As previously mentioned the two controllers implemented are from Skjetne 2014 Two advance control design were implemeted tuned and tested a LgV backstepping and a Nonlinear PID designed controller Only ellipse path was used in the laboratory This had to do with space constraints and a wish to have long run time on the experiments The gradient optimization finds the fastest or steppest change this helps the controller converge faster toward the desired setpoint An unfiltered update law can be sensitive to noise in the measurment while a filtered update law woould smoothen out and therefore be more stable with noisy measurments However this will i
64. terature including pipe laying operations transit operations and cooperative formation control for groups of surface vessels Recent developments have resulted in a generalization of the original maneuvering problem Instead of focusing solely on one dimensional paths the objective is to ensure that the output of the controlled system converges to any desired manifold This extension provides greater flexibility effectively extending the possible applications of the design methodology Although several applications already have been documented for marine vessels such as formation control Line Of Sight LOS based guidance and control and extensions of maneuvering based path following with positional constraints few experiments have yet been conducted The aim of this thesis is therefore to design a LOS based maneuvering control law for the model ship C S Enterprise I CSE1 implement this on its real time control system architecture and test it in the Marine Cybernetics Laboratory MC Lab Work description 1 Allocate time in MC Lab for experimental testing 2 Describe the new modularized HW SW architecture for CSE1 This should show how different control modes for CSEI is implemented in separate modules and how one can switch between these control modes The description should explain the main function s of each control mode and what resources that are required e g what measurements must be available communication etc 3 Perform a
65. terprise 1 Matrices LL eee 25 2 0 8 5 25539 RR se ee we pees 25 3 System identification 27 WA ee eG a ee Rhee he deg eee LL Rod dm oon RY 28 Thruster mappIBE suoi kw ee AA A dem Sop A ARN E BCE 28 4 Line of Sight Experiments 33 4 1 Simulation FUNS e e 444452 o momo Rub Rb EAR 9 Ee ES 33 12 Laboratory TONE n he By Diels 33 5 Conclusion 35 DE PHUUEG VORS uou ae ee ee d 36 Bibliography 36 Glossary 39 Acronyms 41 Symbols 43 Appendix 44 A Thruster plots 45 B Draft for User Manual 73 Chapter 1 Introduction This chapter provides the general background information to grasp a better understanding of the following chapters The thesis can be shortly described as a report focusing on the maneuvering of a surface vessel in calm water and slow speed and the infrastructure surrounding CyberShip Enterprise 1 CSE1 1 1 Motivation In the academic world of marine cybernetics there are many great ideas However they remain there due to seveal factors among others the amount of resources need to process refine simulate and verify them Those that makes it past those initial stages often stops on the boarder between the academic and real world The reasons can be many one might be limited possibility for real world verification Real world experiments are an important part of verifying designs and theories The cost of doing this i
66. th of them were fully implemented LgV backstepping proved to handle the uncertainties better Although it would overshoot when exiting the sharper part of the ellipse path The vessel is able to follow the virtul point it is chasing but the virtual point is unable to stick to the desired path While the fully implemented Nonlinear PID would struggle to converge to the path and behave erratic The velocity dependent term distorted Nonlinear PID s results the most The reason is most likely due to the quality of the estimations of the velocities from the observer The feedfoward term also deteriorate the control and prevents it from converging If the Nonlinear PID was partially implemented deactivating the feedforward and the velocity dependent term then it is the superior one It would converge naturally to the desired position Once on it would stay there indefinitely despite of the broken servo arm Chapter 5 Conclusion The reliability when conducting laboratory experiments with CyberShip Enterprise 1 have greatly improved This was achieved through repositioning and replacing the antenna for wireless communication and a restructuring of how signals were handled between servers and softwares The original setup with single frequency was replaced with a multi frequency producer consumer structure The futherst away a vessel is visible for Qualinsys is around 17 meters from the cameras The shortest distance is roughly 4 meters infront o
67. thuster two close to Main battery one on each side for each Voith Schneider propeller The indicators on ACT LiNK port 1 should light up green to indicate communication with HILlab Test communication opening Command Promt write ping 192 168 0 77 Command promt should show something like C Documents and Settings mcl gt when opened and C Documents and Settings mcl gt ping 102 168 0 77 Pinging 192 168 0 77 bytes 32 time 5ms TTL 64 Pinging 192 168 0 77 bytes 32 time 5ms TTL 64 Pinging 192 168 0 77 bytes 32 time 5ms TTL 64 Pinging 192 168 0 77 bytes 32 time 2ms TTL 64 Ping statistics for 192 168 0 77 Packets Sent 4 Received 4 Lost 0 lt 0 loss gt Approximate round trip times in milli seconds Minimum 2ms Maximum 5ms Average 4ms after a successful ping The most imprtant thing is that you receive packets in return the time might vary but the important thing is that it responds to the ping If Lost 100 meaning no repons means either Laptop or CompactRIO is unable to communicate with HILlab Check Laptop is connected to wireless network HILlab if not connect to it HILlab 79 Check ACT LiNK port 1 are showing activity e g are lit blinking if not check ethernet cable is connected to ACT LiNK port 1 and to the D Link Wireless Bridge if not connect to those Battery gives power to CompactRIO and D Link lights indicators are lit blinking if not
68. till new data is available e g will act as a zero order hold model For one body tracking use QTMDriver_OneBody vi For two body tracking use QTMDriver TwoBodies vi For three or more body tracking 1 Make a copy of QTMDriver OneBody vi or QTMDriver_TwoBodies vi in the same folder 2 Rename it to an appropriate name e g for tracking three bodies QTMDriver_ThreeBodies vi 3 Open it and go to Block diagram 4 Inside While Loop go to Case Structure gt Case Read 5 In there is another Case Structure To the right of it is a Array Subset with 2 constant block attached to it One of those is 0 to indicate first index the other one is 2 number of bodies 7 input the correct number in the second one 79 that determines the length of the array e g one body 9 two bodies 16 3 bodies 23 etc 6 exit and save T NB the SIT input block in Simulink mdl file should have the same port size as the length of the array specified in the QTMdriver else it may spill over onto other sit input ports or become suffled in the order they are sent Acquiring QTM data Connect QTMdriver to vi The data from Qualisys is never explicitly mapped anywhere but it is handled by the FPGA file selected in the hardware mapping e g sitfpga cRI0 9113 IO CSE lvbitx After Connecting the LabVIEW vi file TemplateLV HMI vi to the mdl file TemplateSL mdl via the dll file TemplateSL dll created through Realtime Workshop
69. to y of linear path in O frame in meters m 17 yq Desired position in O frame in meters m 44 Appendix A Thruster plots 45 46 APPENDIX A THRUSTER PLOTS Force N Seperate surge force generation of port and starboard VSP eg f amp Starboard f 4 0 8 06 0 4 02 0 02 04 06 08 1 u value setting Figure A 1 Measurments of VSP speed at 0 4 coefficient of CSE1 47 Force N Superposition of seperate surge force compared to simultaneous surge foce S Superposition of f and f Simultaneous of f s 1 08 06 04 02 0 02 04 06 08 1 u value setting Figure A 2 Measurments of VSP speed at 0 4 coefficient of CSE1 48 APPENDIX A THRUSTER PLOTS Force N Seperate sway force generation of port and starboard YSP e Port t Starboard fa 0 8 06 04 02 0 02 04 06 08 1 u value setting Figure A 3 Measurments of VSP speed at 0 4 coefficient of CSE1 49 Force N Superposition of seperate sway force compared to simultaneous sway force S Superposition of f and f 4 Simultaneous of f a 4 0 8 06 04 02 0 02 04 O06 08 1 u value setting Figure A 4 Measurments of VSP speed at 0 4 coefficient of CSE1 50 APPENDIX A THRUSTER PLOTS Force N smoothing of f smoothing 4 0 8 06 04 02 0 02 O04 O06 08 1 u value setting Figure A 5 Measurments of VSP speed at 0 4 coefficient of CSE1
70. trl z or Edit Undo Window Move to undo it O ONLY delete all wires inside Init Code Read Code and Close Code Flat Sequence Structure Gray frames below the name tags 1 Hover above Ring port blue top left gt right click gt Create gt Constant 2 Repeat 1 for all added vi blocks 3 For each constant change it to appropriate Value by left click Constant and select correct value For vi block in Init Code Flat Sequence Stucture Constant Init For vi block in Read Code Flat Sequence Stucture Constant Read For vi block in Close Code Flat Sequence Stucture Constant Close 4 For all Flat Sequence Structure with vi blocks connect Error wire alternating yellow black wire to it left Error port to error in port error out port to right Error port 5 From Init Code Flat Sequence Structure vi block port to Read Code Flat Sequence Structure vi block port Connection ID out to Connection ID Turquoise wire QTMqueue Out to QTMqueue Orange with turquoise shell wire VI Refnumb QTMTask Out to VI Refnumb QTMTask light turquoise wire 6 Read Code Flat Sequence Structure vi block Hover above DataQutIndex port right click gt Create gt Constant and set Constant Connect Flat Sequence Structure presized input array Orange wire port to vi block left presized input array port to DataIn port Data0ut p
71. ve the strongest coupling in surge sway Two advance control design were implemeted tuned and tested a LgV backstepping and a Nonlinear PID designed controller In the ideal simulated word both of them were equal in terms of maneuvering Only ellipse path was used in the laboratory This had to do with space constraints and a wish to have long run time on the experiments Originally only the LgV backstepping was tested in the laboratory It converged and per formed well The only downside was it would constantly overshoot the heading resulting ii iv Preface in a constant oscillation while moving alone the path The reason was due to the noisy velocity estimation In the laboratory their performance depended heavily on how they were implemented If both of them were fully implemented LgV backstepping proved to handle the uncertainties better Although it would overshoot when exiting the sharper part of the ellipse path While the fully implemented Nonlinear PID would struggle to converge to the path and behave erratic The velocity dependent term distorted Nonlinear PID s results the most The Nonlinear PID is the superior one if only the first term iss active It would converge naturally to the desired position Once on it would stay there indefinitely despite of the broken servo arm With all the improvement in the reliability one important problem still remains the loss of visibility There are incidents when Qualinsys disp

Icon - To Daim

Contents

Download Pdf Manuals

Related Search

Related Contents