Home

- Murdoch Research Repository

Contents

1. Mr Bruce Johnson BEC Engineering Mr Simon Dreyer BEC Engineering Mr Adam McConkey BEC Engineering Professor Parisa Bahri Murdoch University Associate Professor Graeme Cole Murdoch University Doctor Gareth Lee Murdoch University Audrey West Table of Contents Declaration EE 1 A ege E 2 Aen E een E EE 3 Table Ol e 5 rate cette aa sete a a a a aA a LE an raaa EEEE RRA 4 Ae OP Nai ege eler Le EA A AAA 6 1 1 te e tel e DEE 6 EE 7 Tg CODE e 8 WANG ONCE EE 9 1 5 System development and testing EE 10 SEENEN 12 1 7 Project Team Member ce sccec ta veeessliseteee ca peeneevasuiees apecsseveeesassctpeeeeeeree 13 Se SCANMGN sites ee ee 15 2 1 Theory of operation EE 15 2 2 Data Communications dees eieiei re ege 16 Page e NEE 17 2 4 Retrieving scan Values Nee AEN EEN 19 2 5 Send re 20 2 6 Receive Telegrams 2 cece cee ccve ede ccue ech cccie ee Eege 21 2 7 Turning the laser on and Be EE 22 310 Embedded EE 22 3 1 Hardware and operating system 22 3 2 Processing BEE 23 Foes at ie Le de EE 26 3 4 le elle ek Ee 26 3 5 TOME ane 31 2D Me ien EE 33 SO LS ORE NEE 33 SRA E AV S tices Ee ee esterase wad Ee Ee ENEE 34 CABET oz ols eee ee ee 35 3 7 1 Database Software Functons 37 32 2 VIEW database e TE 38 3 8 Alarmi LOOP WEE 39 39 PLC C MM UNICA ONS meesi eanan ates a ite ete eke 39 3 10 Eror e Ein ell Le DEE A0 AO SOA TR E 41 4 1 SCADA CommMm NICAUONS sirrinin a aE a a Er naak 41 EL EE 42 4 3 Buffering scan data E 45 4 4 eine
2. 10 car rear alignment A third table titled scandata contains arrays of raw polar co ordinates for each scan where the ore car triggered an alarm This information is necessary so that if operation staff wishes to visually see the car which triggered an alarm the information can be posted onto the SCADA network Each individual scan is kept as a separate record with a field point_num that contains an ordinal value for each record The records are grouped by another field named time which contains a string with the date and time the alarm event occurred The time value is used to uniquely identify one car from another and used to retrieve a recordset of scan result records when requested The time value was used because it is the only variable unique to each alarm event The car ordinal value is duplicated when a subsequent train passes through the dumper The records are placed in the correct order using the field point_num 36 The fourth table stores the system set point values This table only has two fields which are titled tag name and value The tag name is simply a name for the variable as a string All the values are of type double These values are stored in a database instead of being stored as local variables in RAM This means that these values are stored on the hard drive so that the values can be retained when the system is restarted or if the power to the system is lost These values were kept in a database format in order to keep th
3. This calculation assumes that there is always a full spot strike SICK 2006a In the event of an edge strike the raw result will be scaled incorrectly and an incorrect measurement will result The measurement could be either closer than the object or further away SICK 2006a Figure 3 6 shows a cross sectional profile which contains a co ordinate affected by edge strike These edge strike values are random in nature and a similar effect on the system as noise from any instrument An algorithm had to be developed which made the calculation results immune to the effects of this random value The starting and finishing points must be inside any points affected by edge strike and be consistently reproducible Point affected by edge strike Figure 3 6 Cross Sectional profile 29 Three algorithms were developed and tested for consistency and immunity to edge strike The first algorithm tested searched for the top of the side walls within a fixed window as shown in figure 3 7 The algorithm found the mean X co ordinate for all the Y co ordinates above a predetermined set point over the entire length of the car This gave an X value representing the top and centre of the wall which could be consistently reproduced from one car to the next to be used for a reference and was also relatively immune to the effects of edge strike Finding an X co ordinate representing the centre of the car is now a simple mat
4. At the time this report was written installation of the equipment on the prototype site had been completed but testing and commissioning had not yet begun The software had been developed using the cardboard models and software testing had progressed significantly The software was relatively stable and met the functional specifications but still had some minor bugs 5 2 Scope of work changes During the development of the laser scanning system the ability of the laser scanner started to become apparent At the request of the client testing would be moved from the workshop in Perth to the clients workshop located near the installation site The relocation of the testing would enable testing to be carried out using real ore cars rather than cardboard ones In this way the system could be effectively tested on real ore cars without disruption to the working car dumper Figure 5 1 shows the proposed location for testing the laser system The laser scanner will be mounted from the hand rail on the walkway above the canopy A single car can be pushed by hand through and under the scanner to simulate the car being ejected from the dumper A description of the tests to be performed in this workshop is contained in appendix xi Figure 5 1 Proposed Workshop test Site 50 5 3 Future Work The immediate goal for this project is to complete all the work which has been approved to proceed by the client This includes the testing of the laser syste
5. and used by Drexel n d This algorithm used the command 20h 24h to request measured values continuously This command will cause the scanner to send arrays of data one after the other without waiting for a request The interval between scans depends on the baud rate selected The scanner will send the next telegram as soon as the previous telegram is finished Using the continuous mode means that all scans can be retrieved but the scan rate cannot be controlled At 500 000 baud the interval between telegrams is approximately 75 milliseconds This project requires a scan interval of approximately 200 milliseconds therefore we must discard every second or third scan The variation in scan intervals will cause an error in the integration calculations however a high degree of accuracy is not required in this application To achieve greater accuracy there is acommand which will return the measured values with a real time index This index could be used as the third dimension for real time scanning applications 20 2 6 Receive Telegrams The structure of the receive telegram is shown in figure 2 4 To be able to convert the received signal to a set of values the telegram has to be dissected A flow chart of this process is shown in figure 2 12 In this project the telegram interpretation is achieved using the algorithm developed by Rodger 2007 This algorithm is a stable and industrial strength method in taking the results from the raw data This
6. on gravity to hold the body down in its correct position In order to test if the scanner will detect the event that a car is not on the bogie correctly the car will be jacked up and chocks placed between the body and the bogie to simulate the situation where the body is not correctly sitting on the bogies The output level value is the mean of the difference between the side walls heights measured in centimetres at the points shown in figure 3 9 The ore car bodies are very rigid and not prone to twisting Variation of level from one end to the other is very small 34 LEVEL Figure 3 9 Car level 3 7 Database Included in the scope of works for this project was the facility to store scan results in the form of a database Initially this database will be stored on the embedded PC and may be moved to a file server in the future The local database contains four tables The first table contains data for each type of ore car The second table contains the results for each car scanned while the third table contains raw scan results for any car which tripped an alarm The fourth database will store set point and other variables All results data will be kept for 60 days before being deleted automatically The SCADA system however will only have access to data on a train while it is being unloaded 35 Table one titled profiles contains information about each type of ore car that is owned by the client The ength field is the measurement o
7. ordinates together with an index value for the order of the scans in the car and the time that the alarm was generated The function ShowAlarmResults extracts the results from the database and loads them onto an Active X SCADA communications object The Active X object emulates a PLC and is discussed in detail in section 4 1 This function uses the time the alarm occurred to execute an SQL search to find the correct record The results are directly loaded onto the registers to be read by the SCADA This function is discussed further in section 4 4 37 The ShowAlarmData function retrieves the polar co ordinates from the database This function uses an SQL search to find a recordset which contain a set of scans for an ore car which set an alarm A recordset is found using the date and time the alarm occurred as the unique identity Records are placed in correct order by using their ordinal value Values are placed onto the SCADA Active X registers six arrays at a time The function keeps track of which arrays have been placed for collection by using a pointer Each time the function is called it sends the next six polar arrays The function is called by CheckAlarm at intervals from the timer if the last arrays have been collected and there is more arrays to be posted Every time a train has finished being unloaded the cleanup function is invoked This function examines the results table and scan data table The tables are searched to find r
8. 4 5 6 ton N 1 amp n 2 Position Description Startof Address length Response Data Checksum text Table 2 11 Response Telegram structure SICK 2006a 19 2 5 Send telegrams Two different strategies were assessed for retrieving scan data from the laser scanner The first strategy is based upon Rodgers 2007 who used the command 30h 01h to request a single array of scan measurements This telegram is constructed as a string and sent to the scanner using the MSCOMM32 Active X component included as part of the visual basic package Using the visual basic event timer the telegram is sent at regular time intervals The response from the scanner is one telegram containing an array of measurements The algorithm developed by Rodgers 2007 was for 9600 baud and failed to work properly at 500 000 baud The software on the embedded PC waits until it receives an array of scan measurements from the laser scanner once a request for measurements has been made Sometimes there was no response from the scanner This problem was investigated using an oscilloscope which showed that the command was simply ignored by the laser scanner The send telegram contains a checksum used by the scanner to test the integrity of the telegram SICK 2006a It is likely that the data was corrupt however this could not be proved The second strategy used was to retrieve the measured values based on the algorithm discussed in SICK 2002
9. 48 5 0 Project Management irnia Ee e eren Ear 49 ae PTOGUGS Safa eek ler ese a eaae e ee a ee 49 5 2 Scope of work EE eege ann ae eg 50 eS ee 51 SO SUNT AY eege 52 PIGIOT CINGS Scitech cect tep heats e EE Ee 54 kleine 56 Appendix i Notes to the accompanying CD ROM 56 Appendix ii Procedure for changing scanner baud rate eee eee 57 Appendix iii Sample of raw test data for finding the side ceeeee 58 Appendix iv embedded PC calculation flow chat 59 Appendix v sample of raw test data finding the aide 60 Appendix vi Sample of database results table 62 Appendix ATTEN 63 Appendix ix Citect tag TEE 64 Appendix x sample of raw data test Citect communicatons 66 Appendix xi Site acceptance test procedure EE 67 1 0 Introduction The aim of this report is to discuss the development of a system using laser scanning technology The motivation for this project is to reduce the occurrences of train derailments caused by unbalanced loads in open rolling stock A SICK laser scanner is to be used to examine the rolling stock after having been emptied Data from the laser scanner will be interpreted by an embedded PC which is interfaced into existing control system infrastructure The development of this concept has been made using models of rolling stock in a Perth workshop by BEC engineering 1 1 Background In some mining operations ore is transported by rail The ore is transported b
10. LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 4 LHside1 dy 7 LHside2 y 7 LHside3 dy 3 LHside4 dy dx 7 LHsided dy dx 3 LHside1 dy 6 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHsided dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 5 LHside4 dy dx 7 LHside5 dy dx 5 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHsided dy dx 5 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 5 LHside4 dy dx 7 LHside5 dy dx 5 58 Appendix iv embedded PC calculation flow chart The flow chart shows the logical flow for the calculations made by the embedded PC as described in section 3 Wait 100mS Start scan Enable clock Scan wagon and record profile Scale to head angle Convert to Cartesian Discard results outside limit Subtract height Integrate LHS scan Integrate RHS scan Finished car Increment scan counter Calculate difference in integrals Calculate Total Volume amp Tonnage Calculate difference in side locations Calculate difference in side heights Reset integrals Compare calculations with set points amp alarm if necessary Send Data to SCADA Set Finished Stop LMS Close Dbase Appendix v sampl
11. a new car when all the registers and the GUI are being refreshed For this application 1 in 200 is an acceptable loss in comparison to the first algorithm and better than the first algorithm which lost approximately 25 percent of the scans 47 4 4 Viewing historical results Included in the scope of works for this project is the requirement to view previous alarms A Citect SCADA operator must be able to read historical information from the database In the event that an alarm event is generated the results and chart will be saved to the database for access by the operator This is necessary because the operator may be watching other screens beside the graphing screen The SCADA screen displays the number of cars which have tripped alarms for the train which is currently being unloaded At the start of the next train this value is reset to zero The operator can view any car which triggers an alarm by entering the alarm number into the field and pressing the enter key This event will set the alarm view flag in the MODBUS slave registers and the alarm number entered onto the appropriate MODBUS slave register The flag will tell the embedded PC to go into historical mode which will search the database on the embedded PC for both the results and chart plot data of the given alarm number These values will then be placed on to the same tags used for the live scanning process which is then displayed on the SCADA screen The operator can then able to se
12. algorithm periodically polls the Active X object MSCOMM32 and extracts the data from the objects buffer A search is made through this data for the telegram header and any unwanted data is discarded leaving an array of polar co ordinate results An alternative method of interpreting receive telegrams can be seen in Drexel n d This method is based on the same general flow chart as Rodger 2007 but uses the event generator in MSCOMM82 When the MSCOMM3z2 buffer size exceeds a preset threshold the object will call an event function Prepare data structure for LMS data capture Send commands to set required measurement mode Start continuous LMS data output Parse incoming data stream for output string header Header found Evaluate length bytes Buffer lt length gt LMS data points in data structure n Parallel task Stop continuous LMS data output Process filter display evaluate LMS data End Figure 2 12 Flow chart for receive telegram interpretation SICK 2002 21 2 7 Turning the laser on and off In order to preserve laser life the laser will be turned off when there is no train being unloaded The electronics will be left on permanently to prevent condensation forming on the laser window The laser scanner also takes approximately 2 hours to reach thermal equilibrium after it has been turned on SICK 2006b Turning the laser on and off can be a
13. ordinates relative to ground level This height value is also read from the text based configuration file 23 Figure 3 2 Scanner calibration parameters tilt angle and height The laser scanner measures all the points over a radial arc of one hundred degrees In most installations the scanner is located well above the height of any ore cars and therefore only a small part of the arc is actually used for the measurements To ensure the integrity of the calculations minimizing computational effort and restricting the field of view available in the SCADA HMI the unwanted results must be removed A typical installation will only require measurements over a thirty degree arc In order to filter unwanted data the software requires an arc size in degrees for the scanning window as seen in figure 3 3 This value is normally large enough to see the whole ore car with a bit extra to allow for variations in the cars This scan window angle is also read from the text based configuration file when the program is started 24 SCANNING WINDOW Figure 3 3 arc angle required to scan the ore car The next step is to convert the polar co ordinates to a set of Cartesian co ordinates The polar co ordinates are contained in an array where each value represents the length from the scanner to an object The polar angle is implicit where the angle value is taken as the array subscript noting that these values are all measured 0 5 degrees apa
14. to CSA Dun ec SC Za 31 A second integration is then performed to find the volume In this case we introduce the third dimension The scans are taken at regular time intervals as the ore car passes underneath the scanner If a constant velocity is assumed then this third dimension can represent the approximate Z co ordinates Because there is no definite start and finish point we can no longer use the trapezium rule Instead we have to divide the length of the car into arbitrary AZ values and assume that the scan has been taken in the middle of each division This method is the same concept as the midpoint rule for integration Stewart 1995 The AZ value is determined from the known length of an ore car divided by the number of scans taken over the car being scanned Using the triggering algorithm detailed in section 3 3 the number of scans taken for each car that passes is counted by the embedded P C and is not assumed to be equal The only assumption that could cause incorrect volumes is the assumption that velocity is always constant The actual velocity value is not required however increased velocity will result in less scans per car and lower resolution The length of the ore car is recorded in the database containing the information on each car type and retrieved as required The length divided by the number of scans gives us an equivalent to a AZ value Volume is now calculated to be Volume Carkengih 3 CSA NumberOfScan
15. value of the buffer size at that time These values were saved into a text file for examination appendix x is a sample of this data It was determined that between the time the buffer size was read by Citect and the embedded PC received the acknowledgement from Citect data was added into the buffer Any data placed into the buffers during that time was lost and this explains the missing arrays of values 46 The third algorithm tested made use of the Automated Solutions MODBUS Slave event generator to determine which buffers had been collected by Citect Any buffers not collected were moved to the front of the embedded PC queue for collection on the next cycle This algorithm resulted in the frames getting out of order Investigation of the request file showed that if a new frame needs to be inserted into the buffer while the previous set of frames are still being read then the new frame could end up either side of any uncollected frame This depended on which order the operating system decided to process the instructions The fourth algorithm tested also made use of the MODBUS request event generator to update the registers values The event generator flushed the buffer contents to make room for new data This algorithm resulted in duplication of frames Investigations of the event files showed that Citect does not always request new data from the MODBUS slave when new data is requested by the Citect programming Sometimes the old values ar
16. 0 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 33 PM 10 23 2008 2 57 33 PM 10 23 2008 2 57 33 PM 10 23 2008 2 57 33 PM 10 23 2008 2 57 33 PM 10 23 2008 2 57 33 PM 10 23 2008 2 57 33 PM 10 23 2008 2 57 33 PM 10 23 2008 2 57 33 PM 10 23 2008 2 57 33 PM 10 23 2008 2 57 33 PM 10 23 2008 2 57 33 PM 10 23 2008 2 57 33 PM 10 23 2008 2 57 33 PM 10 23 2008 2 57 33 PM 10 23 2008 2 57 33 PM 10 23 2008 2 57 33 PM 10 23 2008 2 57 33 PM buffer 0 3 true buffer 1 3 true buffer 1 1 true Read 40151 to 40151 Read 40011 to 40030 Read 1to 16 newdata true 5 Write 1 to 1 newdata false 5 Write 3to 3 buffer 1 3 false buffer 1 2 false buffer 1 3 false buffer 11 false buffer 1 3 false buffer 1 3 false buffer 1 3 false buffer 1 3 false Read 40200 to 40319 Read 10001 to 10016 Read 40002 to 40101 Read 1 to 16 Read 40151 to 40151 Read 40400 to 40519 Read 40600 to 40719 Read 40800 to 40919 count 52 buffer size 2 buffer 0 3 true buffer 0 3 true buffer 0 3 true buffer 0 2 true buffer 0 3 true buffer 01 true buffer 0 3 true buffer 1 3 true buffer 11 true Read 41000 to 41119 Read 41200 to 41319 Read 40011 to 40030 Read 10001 to 10016 Read 40002 to 40101 Read 1 to 16 Read 40151 to 40151 Read 40011 to 40030 Read 1 to 16 newdata true 5 Write 1 to 1 newdata false 5 Write 3to 3 buf
17. 0219347 1 3307888009595 PM 95 5 5 2 30353307115 4 643 9 26 2008 1 42 52 o Ja 20 14865513285 7 4795938385044 6 90706224283891 3 1672653235867 0 79773216917 0 5179160780696 PM 12 3 7203 89 9 26 2008 1 43 08 o Ja 22 74579202543 12 717314282519 8 86577657698839 2 5071194357286 1 9824496195320 PM 4 6 2 40416922619 8 193E 02 9 26 2008 1 43 31 fo Ja 10 88137500062 1 3226052971810 2 38969242586012 3 0509950744506 1 4644259297905 PM 15 6 5 1 17864609615 8 621 9 26 2008 1 46 43 o Ja 25 04645423290 10 487636830745 8 88352276591187 3 6397043505392 2 0403270001572 PM 22 3 1 2 37916453229 9 628E 02 9 26 2008 1 47 34 Jo Ja 1260992842357 0 1519684406465 3 1144899957312 3 1904742160545 0 75288199493 1 9921093374563 PM 14 98 2076 8 9 26 2008 1 48 19 Jo Ja 14 76459905163 2 2367311146195 4 25033254156337 3 1319669842536 0 51563488714 1 1519312897449 PM 4 2 2119 7 9 26 2008 1 54 49 o Ja 19 34813459247 9 1419742846591 7 12252721928327 2 5515400769537 E S PM 4 1 1 1 90985444309 4 8133067090674 607E 02 3 9 26 2008 1 55 53 fo Ja 24 50089150180 10 332591952995 8 70837086370068 3 5420748872027 1 42596009907 2 0997930120901 PM 69 8 9 514E 04 9 26 2008 2 00 37 0 Ja 22 71085160667 8 3743432581645 7 77129871620962 3 5841270871273 1 74662206739 1 8118966121449 PM 39 7 4 44 7 9 26 2008 2 02 04 jn Ja 4 8395549357756 5 04083345461759 0 7614441294678 PM 15 32377888269 2 2 6210559867
18. 21 as shown in figure 2 1 was selected and has an arc of 100 degrees With measurement intervals of 0 5 degrees 201 individual measurements are obtained Distance from the scanner to an object is measured using the time of flight principal A pulse of light is transmitted toward an object and the time taken for the light to be reflected back to the unit is measured The return time is directly proportional to the distance between scanner unit and object as shown in figure 2 2 15 T Transmitter R Receiver Figure 2 2 Time of flight principal SICK 2006b There are two modes of measurement available using the SICK scanner The first default mode will give the measurement resolution in millimetres and a measurement range of 8 1 metres The second mode has a resolution of 1 cm and a range of 81 metres In most of the site installations the scanner will be further than 8 metres above the ballast therefore the second mode will need to be used A change in mode or other configuration settings is achieved using the LMS user software contained on the CD that comes packaged with the LMS unit This software is also required to change the baud rate for the LMS scanner A detailed discussion of the selected baud rate used for this project is presented in section 2 3 Configuration settings can be made permanent by writing to the scanner EEPROM Once permanently configured using the LMS user software there does not need to be any provisi
19. 297 1 91537955479 61 48 9 464 Jg 26 2008 2 02 40 Jo Ja 11 93406354487 2 7774116366273 2 28916297706095 3 6778687953746 1 17555337655 1 6579468985962 PM 12 3 2 384E 02 6 9 26 2008 2 30 29 o Ja 21 27385764269 8 0902954648110 7 34103827687726 3 2958905444717 0 84544082766 0 8063548330665 PM 8 1 6 2326 36 9 26 2008 2 32 05 Jo Ja 17 70083208946 3 1085128217661 5 20233622780723 3 6480798169241 1 6264098461464 PM 28 5 5 1 05505704794 6 551E 02 62 Appendix vii Alarm flow chart Shown below is a flow chart displaying the logic behind the system alarm loop This loop is triggered by a two second timer and checks communications with all parts of the system before turning on the scanner Any faults are logged into the event text file Open Dbase Initialise PLC comms Wait 2 seconds Check PLC Comms Message user Set Fault Alarm Shutdown Start amp Check LMS status Open serial Port YES Scanning Check DBase gt SS YES NO Retrieve car No Start Scan 63 Appendix ix Citect tag list Below is a table of all the variables which are passed between the embedded PC and the Citect SCADA system using MODBUS TCP IP over Ethernet This table is a cross reference list and is necessary to be able to configure the tag list in Citect so that Citect knows where to get the information from The MODBUS Register Number refers to the register number of the Active X MODB
20. 500 38400 250 500000 75 Table 2 7 Minimum scan intervals The baud rates 9600 19200 and 38400 are standard baud rates Microsoft 2008 and will work with a standard PC serial communications port To enable serial communications at 500 k baud a specifically manufactured serial port and software capable of communicating at high speed is required 500k baud is much faster than the traditional serial speeds and is also a non standard baud rate Communication at 500k baud requires a serial PCI card that is manufactured to SICK specifications SICK 2006b This project is coded in Visual Basic 6 which uses the Microsoft Active X object MSCOMM32 for serial communications connections MSCOMM32 will only accept serial connections at the standard baud rates 9600 19200 etc up to 115 k Microsoft 2008 To enable serial communication connections at higher speed the SICK specification PCI card has provision to multiply the clock rate Over clocking is achieved by closing the X8 jumper on the PCI card as shown in figure 2 8 This jumper sets the clock speed to 8 times the speed specified by MSCOMM2z2 This allows the MSCOMM32 object to think that a slower clock speed is being used For example with MSCOMM32 baud rate set to 56000 and with 8 times clocking enabled gives a true serial baud rate of 461 000 even though 56000 is an acceptable baud rate to the MSCOMM32 component 17 Figure 2 8 High speed serial Communication PCI card A s
21. 9 254 64101 6 264 63231 7 8 66865521 580 620 451359 373 396488 256 511 085 264 239602 10 6337202 580 620 451359 373 067649 256 511 085 264 242694 10 6337202 580 619 46743 373 067649 256 666924 264 21256 60 Algorithm 1 mean X value x result 279 403 265 41 6 264 429 265 076 265 368 271 583 265 02 265 065 251 1 59 256 401 256 297 250 297 257 1 51 250 554 251 004 250 13 242 1 49 235 862 235 504 235 929 235 766 235 876 235 844 227 872 y1 600 7728 581 8202 592 658 585 7612 589 7023 609 6281 588 717 594 6285 581 6938 610 3926 610 3926 591 5588 607 4368 590 5723 588 5993 599 4508 606 3563 588 6122 600 4643 588 6122 597 5012 600 4643 601 4519 603 4273 y2 621 7689 602 7406 608 6442 608 6442 601 7567 615 8738 612 5799 610 6121 605 4663 619 4674 621 4353 609 4073 619 4674 613 3483 612 3631 621 2303 618 2746 609 3158 612 2753 611 2888 619 1808 617 2078 617 2078 626 0863 y3 625 1748 625 6991 624 7165 624 7165 621 7689 625 1748 621 7689 628 6467 618 4835 632 5768 632 5768 619 4674 621 7689 619 4674 626 3549 627 3389 633 2424 619 2598 628 1271 628 1271 618 2746 628 1271 628 1271 625 1713 y4 608 1307 627 1369 635 9664 627 1369 626 1558 625 1748 614 3831 623 734 631 5943 622 7515 608 0135 610 9611 632 2585 620 4514 620 4514 619 4674 606 6763 606 6763 Algorithm 3 moving point avaerage x result 302 449 284 3
22. 91 282 834 284 391 284 564 284 737 284 737 286 64 270 47 268 99 268 99 269 1 55 270 799 270 634 273 1 01 272 607 254 485 254 641 256 51 1 256 51 1 256 667 258 693 258 693 241 338 y1 284 737 270 1 41 270 306 270 306 270 799 271 785 270 799 269 648 256 823 256 667 256 355 256 667 256 667 256 667 255 576 255 42 242 368 242 221 240 896 240 896 242 368 240 896 240 896 226 821 y2 270 799 259 31 6 258 381 258 381 259 472 258 225 257 758 258 069 244 282 243 546 243 546 243 694 243 988 243 1 05 243 252 241 927 229 593 229 1 78 228 762 228 9 227 791 228 068 228 068 215 459 y3 285 995 271 283 270 507 271 026 271 61 2 271 583 271 098 271 452 257 1 92 256 401 256 297 256 505 257 1 51 256 802 257 309 256 651 242 1 49 242 01 3 242 056 242 1 02 242 275 242 552 242 552 227 872 61 Appendix vi Sample of database results table This chart displays a sample of data taken from the results table of the database after testing The database has been constructed using Microsoft Access and filled using the ADO Active X component Database components have been discussed in detail throughout section 3 7 date time P See Cat Tonnage car_inbalance car_LHS volume car_RHS_volume car_skew car_level 9 26 2008 1 42 33 o Ja 20 80621941983 8 6941593321004 7 37509468798498 3 028015
23. A system The alarm loop scans communication from the laser scanner database PLC and the Citect SCADA system A flow chart in appendix vii shows the alarm loop logic In the event that communications is not received from one of the devices a fault alarm will be generated which can be seen by the operator in the control room For record keeping purposes the nature of the fault is recorded into a text based event log file A sample of this file is in appendix viii 3 9 PLC Communications To integrate the laser scanner system into each of the existing site s communication links had to be established with the existing site PLC s Information required from the PLC s include whether a train was present when the cars were being ejected from the tipping cell and the ordinal number of the car being ejected The PLC s are connected to an Ethernet network for SCADA communications thus providing a straight forward approach to establishing data communications with the embedded PC The PLC s on three of the five proposed sites are Modicon Quantum series and the other two are GEC Rx7i PLC s BEC Engineering 2008b All these PLC s can communicate over an Ethernet network using the MODBUS TCP IP application layer protocol For the embedded PC to communicate with PLC equipment software was required to establish communications using the MODBUS TCP IP protocol This was achieved via a commercial software component from Automated Solutions Au
24. B ports serial ports and facility for battery backup power Advantech 2007 The operating system installed on the embedded PC is Windows XP embedded This is a condensed version of windows XP Advantech 2007 This operating system will fit on the flash memory removing the need for a hard drive Advantech 2007 However a 160Gb hard drive has been fitted for the purpose of storing the database The database does not contain crucial information and the PC does not require any backup facility The PC has also been fitted with the 500mhz Quatech RS422 serial PCI card to communicate with the laser scanner 3 2 Processing the results Once the laser scanner has transmitted a set of polar co ordinates to the embedded PC the values need to be interpreted The algorithm for this interpretation is shown on a flow chart in appendix iv Data is received from the scanner through the serial connections and converted to an array of integers as detailed in section 2 6 The scanner head is not always exactly at right angles to the ground On some sites the scanner has to be tilted to avoid obstructions and to have full line of sight In these cases the scan data needs to be scaled down by the cosine of the tilt angle as shown in figure 3 2 This tilt angle is read from a text based configuration file when the program is started Height of the scanner unit above the ground is required to convert the array of polar co ordinates to the array of Cartesian co
25. Energy amp Control Systems Bu MURDOCH UNIVERSITY S wy PERTH WESTERN AUSTRALIA Engineering Faculty of Minerals and Energy School of Engineering and Energy ORE CAR LASER SCANNER Engineering Internship Project Graham West November 2008 Declaration Except where have indicated the work am submitting in this assignment is my own and has not been submitted for assessment in another course Abstract This report discusses the development of a system using laser scanning technology to examine railway rolling stock The scanner is to be used to examine ore cars after having been emptied Principal motivation for this project is to identify ore cars containing residual ore after being emptied or damaged cars in order to prevent derailments The system is based around a SICK laser scanner mounted above the rails which scans the cars as they pass underneath Data obtained from the scan is processed by an embedded PC This report examines laser scanners in general discusses the calculations made by the embedded PC and discusses the implementation into an existing site The report also addresses the problematic components of the system including a suitable search tool to identify the sides of the car and communication interfaces between the various pieces of equipment Acknowledgements The author would like to thank the following people for their assistance and support in bringing this project and report to fruition
26. Stephen 2007 Automated Ship Loader Perth UWA Thesis SICK AG 2002 Quick Manual for LMS Communications setup Waldkirch SICK AG Available at http www pages drexel edu kws23 tutorials sick sick html Accessed 08 10 2008 SICK AG 2006a Telegrams for configuring and operating the LMS2xx Laser Measurement Systems Waldkirch SICK AG SICK AG 2006b LMS200 21 1 221 291 Laser Measurement Systems Technical Description Waldkirch SICK AG SICK AG 2006c LMSIBS Configuration software for LMS2xx LMI400 Version 4 1 Waldkirch SICK AG Lammert Bies 2008 RS422 Serial Information Available at http www lammertbies nl comm info RS 422 html Accessed 07 12 2008 55 Appendixes Appendix i Notes to the accompanying CD ROM The accompanying CD ROM contains the source code for the embedded PC Citect and the raw test data The embedded PC software was written using Visual Basic 6 In order to correctly run this program a visual basic development kit will be required The software also uses three Active X components to run which need to be installed on the PC as discussed in this report Trial versions of these components can be downloaded from the Gigasoft website Gigasoft 2008 and the Automated solutions website Automated Solutions 2008 The citect source code and backup files are also included This will require Citect version 5 42 in order to run correctly The source code for both the visual basic and citect has also be
27. US slave object in the embedded PC while the tag name is the variable name used in Citect Both ends need to know the data type of each variable This table also contains a comment for the readers reference ORE CAR SCANNER ModBus From OO P ob 12 13 14 40001 40002 40003 40006 40007 40008 40011 40013 40015 40017 40021 40023 40025 40027 40029 Registers To Data Type Boolean Boolean Boolean Boolean Boolean Boolean Boolean boolean boolean boolean integer integer integer integer integer integer double double double double double double double double double TAG LIST Citect Tag Name CARSCAN_SCADA_OK CARSCAN_SYSTEM_OK CARSCAN_NEWDATA_CMD CARSCAN_CARSTART_CMD CARSCAN_CAREND_CMD CARSCAN_DATAREQUEST_CMD CARSCAN_TONNAGE_AL CARSCAN_TONDIFF_AL CARSCAN_LEVEL_AL CARSCAN_SKEW_AL CARSCAN_LIMIT_V CARSCAN_HIGH_V CARSCAN_ANGLE_V CARSCAN_BUFFER_OP CARSCAN_ALARMQTY_OP CARSCAN_ALARMREQ_V CARSCAN_TONNAGE_OP CARSCAN_TONDIFF_OP CARSCAN_LEVEL_OP CARSCAN_SKEW_OP CARSCAN_DENSITY_SP CARSCAN_TONNAGE_SP CARSCAN_TONDIFF_SP CARSCAN_LEVEL_SP CARSCAN_SKEW_SP Comment SCADA System status Scanner System status New Scan Data Available flag Car Start Car End Request for alarm information high tonnage alarm high inbalance alarm car level alarm car skew alarm Scanning window Width Height from ballast to scanner Scanner head angle Size of current Data buffer No of cu
28. am design commission audit and supervise construction of power generation systems and power stations The power distribution team design systems for power transmission and distribution including protection systems harmonic analysis power factor correction and earthing Plant integrity team members conduct assessments for reliability operational requirements statutory requirements safety and risk A process control division designs multivariable control systems advanced process control systems PLC systems and process optimization strategies The automation division develops PLC and SCADA systems Operator terminals telemetry systems and remote operation systems BEC n d The company provides total engineering services to the resource public and industrial sectors They pride themselves on being an innovative and integrative company They take care of their various cliental needs and services Offering a wide range of services and applying their expertise on projects that require design turnkey solutions construction supervision commissioning and or operations support BEC n d BEC has an impressive history of working in the Oil and Gas industry at all levels with operators and contractors In the mining sector BEC has worked on all types of mining projects including gold copper nickel iron ore coal and diamonds BEC handles projects in all types of environments BEC n d Additionally BEC engineering provides specialist enginee
29. as established with a serial RS422 while communications between the embedded PC and both the PLC and SCADA was made using MODBUS TCP IP over Ethernet 52 A three dimensional charting package has been incorporated into the Citect SCADA HMI system to provide a visual aid for operation staff Results from testing found that the volume could be measured to a reasonable level of accuracy Highlighted was the problem of finding a reliable search tool to find the sides of the ore cars The measurement for the heights of the side walls was a noisy measurement but quite reliable when using a moving point average Alignment measurements taken were extremely noisy Even with several filter methods tested the measurement could only guarantee to an accuracy equivalent to one degree in the polar plane 53 References Automated Solutions n d ASMBTCP ActiveX Help r2 7 Santa Rosa Automated Solutions Inc Automated Solutions n d Automated Solutions Modbus Slave Active X Control Help r1 6 Santa Rosa Automated Solutions Inc Automated Solutions 2008 Software Solutions for HMI and SCADA Developers Available at http Awww automatedsolutions com products modbus asp accessed 8 9 2008 Advantech 2007 UNO 3072 Embedded automation computer user manual Advantech BEC Engineering 2008a Minutes of client meeting South Perth BEC Engineering 2008b Empty Wagon Detection Scanner South Perth BEC Engineering n d Management Structure Availa
30. ble at http www bec engineering com au BECProfile cfm mode displayManagementStructure Accessed 12 11 2008 Citect Pty Ltd 2008 What is SCADA htto www citect com index php option com_ content amp task view amp id 40 amp ltemid 44 Accessed 04 11 08 Citect Pty Ltd 2003 Citect HMI SCADA Help Topics Connel John 1998 Beginning Visual Basic 6 Database Programming Birmingham Wrox Press Drexel n d Interfacing with the SICK LMS200 Philadelphia Drexel Available at http www pages drexel edu kws23 tutorials sick sick htm Accessed 08 10 2008 Eliason Alan and Ryan Malarkey 1999 Visual Basic 6 Environment Programming and Applications Indianapolis Que Education and Training Gigasoft 2008 Net Chart Control Reference Available at http www gigasoft com netchart controlcomponentslibrary htm Accessed 11th August 2008 lon Tecnnology n d Modicon Modbus Available at http www eliteas com pml PMLACFILES Modbus_6200 pdf Accessed 28 10 2008 54 Microsoft 2008 Visual Basic MS Comm Control Available at http msdn microsoft com en us library aa259393 aspx Accessed 3rd October 2008 Odgen Frank amp Fred Cohn Moving up to industrial Ethernet MODBUS TCP The Industrial Ethernet book Iss 45 Available at http ethernet industrial networking com articles articledisplay asp id 2040 Accessed 28 10 2008 Petroutsos Evangelos 1998 Mastering Visual Basic 6 Alameda Sybex Rodgers
31. chieved using the data communications connection Command 68h 01h is used to turn the laser on while command 68h 00h can be used to turn the laser off 3 0 Embedded PC Raw measurement data is passed from the laser scanner to the embedded PC where the data is processed The raw measurements are an array of polar co ordinates which first have to be converted to cartesian co ordinates Search algorithms have been developed to find the ends and sides of the wagon Integration of each cross sectional profile determines the cross sectional area The array of cross sectional areas is then integrated to find volume The embedded PC also contains databases for data on each type of wagon results raw scan data and a table of set points 3 1 Hardware and operating system Figure 3 1 embedded PC with RS422 serial PCI card 22 At the heart of the system an embedded PC has been used to perform all the necessary calculations This PC will interface between the laser scanner existing PLC s and Citect SCADA systems as previously shown in figure 1 2 The PC will also be used to store the scan records as a database An embedded PC was chosen in preference to a PLC due to the complexity of the calculations and the need for a database This rugged PC has been chosen to suit an industrial environment and has a low power 900 Mhz Intel Celeron Laptop processor It has most of the general features as a standard desktop PC a low power graphics accelerator US
32. d Z points are also established in this function The number of X points is twice and scanning window angle as discussed in section 3 2 because the points are 0 5 degrees apart The number varies between sites and is gained from an input argument to the function Currently the number of Z points has been fixed at 50 The plot values are stored using a 2D matrix for each of the X Y and Z axes All matrixes are 50 by twice the scanning window All of the initial Y values are set to zero which will result in a blank screen 43 The update function is called whenever a new data flag is set by the embedded PC in one of the MODBUS slave registers The Citect system communicates with the embedded PC using MODBUS TCP protocol Citect only has the ability to act as the master so there is no way to generate a new data event from the embedded PC Citect scans the flag at regular time intervals to check for new data This poll rate can be slower than the rate that data is coming from the laser scanner Therefore it becomes necessary to have a buffer which can store several sets of scan results The size of the buffer is set in one of the MODBUS slave registers by the embedded PC The embedded PC will increment this value if the previous scan results have not been collected by Citect Citect acknowledges collection of new data by resetting the new data flag The update function accepts arrays of new X and Y values Arrays are loaded into the chart using one
33. d decision on whether to re tip an ore car Presently the use of CCTV is difficult during inclement weather and the evening shifts BEC 2008b An objective for the longer term is to be able to use this numerical information as part of the control system In the event that the amount of residual ore or degree of imbalance exceeds certain set points the control system could return the car to the tipping cell to empty any residual ore A return to the tipping cell needs to be conditional to the ore car still being on the rails If an ore car is already derailed when it is returned to the tipping cell damage to the tipping cell will result BEC 2008b This project entails development of a system enabling the detection of 1 The quantity of residual ore in the car after the car has left the tipper 2 The extent of which any residual ore would be unevenly distributed 3 If the ore car is on the rails and undamaged after it has left the tipping cell BEC 2008b 1 3 Scope of work The principal instrument used to examine the ore cars will be a laser scanner produced by SICK technologies SICK 2006b in Germany The scanner will be interfaced to an embedded PC which interprets the results and records these into a database In addition part of this project is to interface this system to the existing PLC on site which controls the car dumper The existing on site SCADA system used by operations staff to monitor operations will also need to be reprog
34. d is shown with the result In this sample methods 2 and 4 have the most consistent result LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 5 LHside4 dy dx 7 LHsided dy dx 5 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 6 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHsided dy dx 5 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 5 LHside4 dy dx 7 LHsided dy dx 5 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 5 LHside4 dy dx 7 LHsided dy dx 5 LHside1 dy 6 LHside2 y 7 LHside3 dy 5 LHside4 dy dx 7 LHsided dy dx 5 LHside1 dy 7 LHside2 y 7 LHside3 dy 4 LHside4 dy dx 7 LHsided dy dx 4 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHsided dy
35. d the car itself resulting in damage to the tipping cell An algorithm had to be developed to confirm that a car is aligned within the tipping cell arms before being pulled into the cell 3 6 1 Skew In order to confirm the car is on the rails after it has been emptied an algorithm was developed to determine if the top of the car is parallel to the rails An early alignment algorithm used the same search tool as the integration algorithm to find the centre of the top of the wall The scans over the first quarter of the car were used to find an X co ordinate representing the centre of the car at the front A second value was found for the rear of the car The values were compared and scaled by 4 3 to give a numerical figure representing the difference between the front and rear of the car relative to the track Testing this algorithm found the required level of accuracy could not be obtained This method was also inaccurate where there was a large amount of ore in the car A third problem found was that in the event where the entire car had been derailed the car could still be parallel with the rails but the derailment would not be detected Two changes were made to the alignment algorithm in order to address these shortcomings The first change removed the search for the inside edge thus preventing problems when the car contains ore Instead of being compared to the rear of the car the second change involved this value being compared with a fixed p
36. dx 5 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHsided dy dx 5 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 3 LHside4 dy dx 7 LHsided dy dx 3 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHsided dy dx 5 LHside1 dy 7 LHside2 y 7 LHside3 dy 5 LHside4 dy dx 7 LHsided dy dx 5 LHside1 dy 7 LHside2 y 7 LHside3 dy 5 LHside4 dy dx 7 LHsided dy dx 5 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHsided dy dx 5 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 6 LHside2 y 7 LHside3 dy 3 LHside4 dy dx 7 LHsided dy dx 3 LHside1 dy 6 LHside2 y 7 LHside3 dy 3 LHside4 dy dx 7 LHsided dy dx 3 LHside1 dy 6 LHside2 y 7 LHside3 dy 5 LHside4 dy dx 7 LHsided dy dx 5 LHside1 dy 7 LHside2 y 7 LHside3 dy 5 LHside4 dy dx 7 LHsided dy dx 5 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHsided dy dx 5 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9 LHside1 dy 7 LHside2 y 7 LHside3 dy 9 LHside4 dy dx 7 LHside5 dy dx 9
37. e of raw test data finding the side 2 These tables are part of the raw data for a program used to compare different ways of finding the side of an ore car Algorithm 1 finds the mean algorithm 2 uses linear extrapolation while algorithm 3 uses a moving point average The objective is to find which method produced the most consistent result algorithm 2 extrapolation method set y1 point y2 x1 x1 x result 6 49505392 580 608 130742 434 149006 302 448828 308 606747 31 1973727 580 627 136877 409 37701 6 284 39104 294 277024 20 4227258 580 635 966419 411 373977 282 8341 54 294 521 243 13 5661323 580 627 136877 412 64477 284 39104 294 243987 29 238346 580 626 155816 409 7401 284 564027 294 243093 17 4841857 580 625 174756 411 91 8603 284 737014 294 191 492 10 6275922 580 625 174756 413 1 89396 284 73701 4 294 1 79422 0 83245857 580 614 383093 415 00481 5 286 639874 293 833386 54 1976649 580 623 73401 385 41 5594 270 469935 279 296474 20 8416149 580 632 576803 391 29716 268 99016 279 50208 27 709037 580 632 576803 390 08625 268 99016 279 5161 72 28 6900973 580 631 594271 389 91 3262 269 15458 279 488653 28 6900973 580 621 768945 389 91 3262 270 798774 279 1 87686 11 031012 580 622 751478 393 027033 270 634354 279 1 88045 6 12571052 580 608 01349 393 891 969 273 1 00646 278 722601 5 14465022 580 610 961088 394 064956 272 607387 278 81 4645 385 961142 580 633 242448 31 0 259433 254 4851 77 266 494002 28 3193055 580 632 258518 370 1 0809
38. e of this is that the value could be out of date Citect has its own algorithm to find the most efficient method of retrieving register values from a device This algorithm is referred to as smart blocking in Citect Pty Ltd 2003 The intention of smart blocking is to minimize network traffic only refreshing tag values as they are needed and retrieving them in blocks rather than as individual variables The idea came about because Ethernet packets contain a large quantity of overhead data used for addressing control checksums etc Citect will block tags where the tags are on the same I O device have the same read request instruction and where the registers are within 120 words 240 bytes Citect Pty Ltd 2003 The 240 byte limit is due to the maximum size of a MODBUS Ethernet packet The MODBUS driver inside Citect has its own cache which contains a queue for incoming data requests After a specified time the driver will use smart blocking to group the requests made Then the driver will request multiple data values from the I O device in one instruction Citect Pty Ltd 2003 AMODBUS request contains an instruction eg read the first register affected and the number of consecutive registers to be affected Citect Pty Ltd 2003 In order to understand how this algorithm operated code was written for the embedded PC which used the request from Citect to generate an event This event recorded the registers requested by Citect the time and the
39. e program structure consistent 3 7 1 Database Software Functions Microsoft Access was used to build the database template and structure Visual Basic 6 has a number of tools to work with this type of database This was the motivation behind using visual basic as the programming language The project makes use of the active X data object ADO developed by Microsoft and uses the JET engine This object can be used to access data on a number of types of databases including SQL Server Oracle Excel and internet based ASP Connell 1998 If the database is ever moved to a file server then this code should facilitate the transfer Six of the functions in the visual basic code use the ADO object to communicate with the access database These functions will require changes when the database is moved to the network server A brief description of these functions is given below The function Loadprofile uses an SQL search to find records for the car type This function is called every time a car has been scanned to obtain the length and reference volumes which are returned by the function as a user defined data type object The function saveResults is called each time a car is scanned after the calculations have been made This function creates a new record and saves all the calculation results The function saveScanData is called when a car triggers an alarm This function creates a new record for each scan and saves the array of polar co
40. e reused resulting in the duplicated frame The fifth algorithm that was tested did not rely on the buffer size register A fifth algorithm used indexing to find updated scan data An index value representing the number of times the current car had been scanned has also been included with the scan data This value was placed in the array of polar values at position zero inside the block of 120 to use the Citect smart blocking to our advantage This ensured that the Citect collected this value at the same time as the remainder of the polar array The Citect algorithm was modified to collect all of the buffers but only to display the ones with a valid index number and discard any frames which had already been displayed on the GUI To prove that the new algorithm worked the index value was temporarily displayed on the GUI along with the index value used to track the Z axis This was used to monitor the number of lost scans This algorithm is far better than the others as it did not duplicate any frames inserted frames into the correct order but still dropped approximately two percent of frames The sixth algorithm was an amended version of the previous algorithm This algorithm called the tag values directly from the embedded PC instead of using the values from the local RAM by calling the Cicode functions tagwrite and tagread This algorithm lost approximately one scan in every 200 scans The loss generally occurred at the beginning or the end of
41. e the image of the car and the tonnage imbalance car number alignment and level A discussion of the algorithm to search for the data has been previously explained in the section 3 7 1 The operator can view live scanning by clicking on the resume button or by entering zero in the view alarm box Any invalid entry into the request box will be ignored 48 5 0 Project Management The project fell significantly behind schedule due to several hold ups relating to the on site installation of the equipment These delays meant that the majority of the development was done using the cardboard models instead of the real installation At the request of the client the proof of concept testing is to be carried out in a simulated environment using real cars instead of the actual car dumpers This project still needs to be completed for the final installations but the concept could be applied to other applications beside railways 5 1 Progress At the time this report was compiled the system had been developed to a point where it could be tested on site Initial project planning was to install the equipment on five sites and develop the software as each was being installed The original project plan failed to materialize due to several hold ups that could not be foreseen The first and most significant was a health and safety problem on the site which was nominated for the installation of the prototype On this site an ablution block had a plumbing problem
42. ecords more than sixty days old These records are then deleted permanently from the database in order to keep the database to a manageable size 3 7 2 View database mode When a request for alarm records is received from the SCADA system the embedded PC sets a flag to go into the historic mode In this mode the scanner results are prevented from being placed into the SCADA communications Active X object registers System outputs and graphics will be prevented from being shown on the Citect GUI All other functions will continue normally The laser scanner will continue to examine the ore cars and generate more alarm events if necessary When an alarm event is generated all the scanner results and scan profiles are recorded into the database with the date and time when the alarm was generated The alarm date and time is also stored into a local variable array When alarm results are requested the embedded PC refers to the date and time corresponding to the appropriate alarm number entered into a field by the citect operator The embedded PC executes an SQL search to find the results for the requested alarm Once these are found the results are placed directly into the SCADA Active X communication object registers 38 3 8 Alarm Loop The embedded PC software verifies communications with all other associated equipment at regular time intervals In the event of failure in one of the system components operations staff are notified via the Citect SCAD
43. en saved in Microsoft word format for viewing only The test results are saved in both text files and excel format 56 Appendix ii Procedure for changing scanner baud rate 1 Open LMS LMI user software 2 From sensor menu select interface 3 Set the current baud rate 9600 is the default for a new sensor and comm port number 4 from the file menu select receive If the connection is successful the software will read the scanner s current configuration 5 From the LMS menu select user group or press F8 In the popup select authorized customer and enter the password SICK_LMS Click on login 6 From the measurement technique menu select SICK diagnosis 7 Select OK at the device prompt 8 On the menu under change operating mode select installation mode and click add The installation mode should appear to the right 9 In the menu select the desired baud rate and click add 10 In the menu under Definition perm Baud rate variant select Constant baud rate and click add 11 Click on run to execute the change Examine the bottom screen for any errors and confirmation of the change 12 Close the popup and the LMS LMI user software 57 Appendix iii Sample of raw test data for finding the side Shown below is a sample of raw data obtained from the program used to determine the most reliable method of finding a point on the side wall which could be accurately be reproduced Here the algorithm metho
44. ermine this value is as follows 1 set the value in the database to zero 2 Measure the alignment of a sample car using the laser scanner Note this value can be obtained from the GUI on the embedded PC or from the results table of the LMSdata mdb database 3 The value to be used shall be the mean of at least 10 test results 4 Save the mean in the setpoints table 68 Testing procedures Test for zero tonnage Confirm that an empty car outputs a tonnage value of approximately zero after being pushed under the scanner Test for zero imbalance Confirm that an empty car outputs an imbalance value of approximately zero after being pushed under the scanner Test for zero level Confirm that an un deformed car outputs a level value of approximately zero after being pushed under the scanner Test for zero alignment Confirm that an un deformed car outputs a front alignment value of approximately zero after being pushed under the scanner Test for zero alignment Confirm that an un deformed car outputs a rear alignment value of approximately zero after being pushed under the scanner Test for tonnage Confirm that a car containing ore outputs a tonnage value which represents a fair assessment of the amount of ore which can be seen in the car after being pushed under the scanner Test for Imbalance Confirm that a car containing ore which is predominantly located on one side of the car outputs an imbalance value which represents a fair a
45. ernative method to determine this angle is to adjust the angle parameter on the embedded PC or by physical adjustment of the angle on the scanner unit itself until the ground level as read by the scanner is equal to zero Scanning window The scanning window is the angle in degrees from vertical to the outside of the ore car underneath the scanner Measured values read by the scanner taken outside this window are discarded by the embedded PC and not displayed on the Citect HMI This angle is required to be wide enough to include the top of the side wall of the ore car The value also needs to include a safety margin of 2 degrees to allow for variations between ore cars In the event that these values are different for each side of the ore car the greater value shall be used Level A physical level adjustment is provided with the LMS scanner The required level is relative to the railway iron as distinct from level obtained from gravity The scanner shall be taken as level when the numerical level measurement on the embedded PC GUI or SCADA is equal to zero This value shall be accessed as the mean of at least 10 ore cars 67 Procedure for profiling empty ore cars For each type of ore car which can be emptied on a given installation a database with parameters of each type needs to be developed The database is located on the profiles table contained in the LMSdata mdb file which can be opened using microsoft access Each car type has four f
46. f Y to exceed a negative set point This program was allowed to run for an hour while scanning a stationary object to determine if any of the methods produced a consistent result A sample of this test data is contained in appendix iii While some worked better than others none were consistent enough to be used as a reference point Further investigation revealed the reason for the inconsistency is that the light was hitting the edge of the object and causing an inaccurate distance value This problem is referred to in SICK 2006a as the edge strike or halo effect 27 Figure 3 4 Edge strike halo effect SICK 2006a Even though a laser is being used this laser is of a low intensity type 1 laser and has a very wide diameter spot SICK 2006a Table 3 5 is a table showing the spot diameter at various distances from the scanner gees nm Ces Le ame oa 1 8 1 8 w s s m C a a EE Table 3 5 Laser spot diameter SICK 2006a 28 In the event of an edge strike where only part of the beam hits the object an incorrect measurement will result The algorithm inside the laser scanner adjusts the raw time of flight measurement by a factor according to the intensity of light reflected back from the object SICK 2006a according to Output measured value raw measured value distance correction in accordance with the internal distance table correction in accordance with the internal received energy table SICK 2006a
47. f length inside the car body in centimetres Length is required to calculate the integral of the cross sectional areas as discussed in section 3 5 The fields LHS volume and RHS volume are the cross sectional volume integrals of empty ore cars These values represent the volume of an empty ore car as measured by the scanner and are subtracted from the volume determined by the scanner to give the volume of residual ore The procedure for obtaining this value is to set the database value to zero and then scan multiple ore cars that are known to be empty The system will record the volumes of these ore cars in the results database A suitable average value can then be used as the reference volumes The two values do not have to be identical They will only be expected to be the same only when the laser scanner is exactly level This table also contains a series of X and Y co ordinates These values are for profiling different car types in a future development The second table named results contains the information calculated for each ore car after it has been emptied and scanned A sample of this table is contained in appendix vi The results table contains the following fields 1 date amp time the results were calculated 2 car ordinal within train 3 car type 4 volume of ore in left side 5 volume of ore in right side 6 the tonnage of the residual ore 7 the difference in tonnage between each half 8 car level 9 car front alignment
48. fer 1 3 false buffer 1 2 false buffer 1 3 false buffer 1 1 false buffer 1 3 false buffer 1 3 false Read 40200 to 40319 Read 10001 to 10016 Read 40002 to 40101 Read 1 to 16 Read 40151 to 40151 Read 40400 to 40519 count 54 buffer size 2 buffer 0 3 true buffer 0 3 true buffer 0 3 true buffer 0 3 true buffer 0 3 true buffer 02 true buffer 0 3 true Read 40600 to 40719 Read 40800 to 40919 Read 40011 to 40030 Read 41000 to 41119 Read 41200 to 41319 Read 10001 to 10016 Read 40011 to 40030 66 Appendix xi Site acceptance test procedure A formal part of testing the laser scanner system includes Proof of concept testing These steps will describe the technical steps to set up and test the system on the testing site and form part of the formal test document Scanner Calibration procedure Height The height of the scanner is to be measured from the ground level to the scanner Ground level is taken to be the surface level where the finished surface beneath the scanner is concrete The height of the scanner above the ground is to be measured to the base of the LMS scanner unit using a tape measure This value is to be entered into the Citect setup page Confirm this value is correctly read by the embedded PC Head angle The head angle is taken to be the angle in degrees of the scanning plane to vertical This angle can be measured as the angle of the scanner base away from true horizontal An alt
49. ields which requiring entries Car type The car type can be any string value used to commonly identify the class of ore Car Length The length of the ore car is to be taken as the internal length of the car This length shall be determined from either mechanical drawings or measured by physical means and using a suitable measuring instrument The measurement shall be taken at a point 1100 mm below the top of the car Left hand side volume The value of the left hand side volume is the volume of half an empty ore car as measured by the laser scanner In the same way that tare is the measured mass of an empty vehicle a volume is required that represents an empty ore car The procedure to determine this value is as follows 1 set the value in the database to zero 2 Measure the volume of a sample car using the laser scanner Note this value can be obtained from the GUI on the embedded PC or from the results table of the LMSdata mdb database 3 The value to be used shall be the mean of at least 10 test results 4 Save the mean in the profiles table Right hand side volume As per left hand side volume Procedure for setting alignment offset Alignment offset is the distance between the centre of the railway irons and the centre of the laser scanner as seen from true vertical This value is necessary to calculate the alignment of the ore cars The value is stored in the set points table of the LMSdata mdb database The procedure to det
50. ill examine the ore cars SICK 2006b The scanner measures distances to multiple points through a radial arc The scanner is mounted above the ore cars so that the arc is at right angles to the car as shown in figure 1 3 The cross sectional profile is scanned several times as the ore car passes underneath The scanner output is a set of polar co ordinates which are converted to Cartesian co ordinates and then integrated to determine the cross sectional area of the ore Car Figure 1 3 Measuring cross sectional profile As the ore car moves along the railway line a scan can be taken at regular time intervals giving a set of cross sectional area values By integrating the array of cross sectional area values the volume can be determined To obtain the volume of residual ore the integral can be compared with the integral of a known empty ore car The difference is found in the same manner as using integration to find the area between two curves The volume of residual ore in both the left hand side and right hand side of the cars are calculated as two separate values The sum of these gives the total volume while the difference gives the imbalance between the two halves Both these values are scaled by a density value taken from the SCADA system to give the total residual and imbalance quantities in terms of mass 1 5 System development and testing This project has been developed and tested using cardboard models The laser scanner was bo
51. ill required significant development time to reach a point where it could be used reliably The BEC Company also now has a benchmark for future projects should there be any requirement to incorporate an embedded PC into a control system that contains standard PLC s and SCADA systems This could be used as a future template for applications in automation robotics or advanced process control where standard PLC s do not have the required computational power and an embedded PC could be incorporated into control infrastructure 51 6 0 Summary This report discusses the use of a laser scanner to examine the contents of railway rolling stock after having been emptied The purpose of the scanning was to determine the amount of ore which may remain in the cars to compare the amount of ore on one side with the other side and finally to determine the alignment of the car The laser scanner output measurements are fed into an embedded PC which also takes information from a PLC to calculate numerical quantities for the residual ore These calculation results are able to be displayed on a SCADA workstation using Citect as a three dimensional chart displaying the profile of the examined car The underpinning principal motivation for this project is to enable reduction of train derailments As residual ore in cars has been the cause of previous derailments This system will benefit all client stakeholders by primarily having a safer working environment for
52. ing an event in Citect which can be used to trigger charting functions For Citect to determine if there is new data Citect must poll the embedded PC at regular intervals and download any available data Because the poll time is slower than the scan rate provision for downloading several scans at once must be catered for 45 Several buffering algorithms had to be trialed to get the transmission of this data to an acceptable level of operation The first algorithm that was tested monitored the new data flag and created an event when this flag was set This algorithm used the Citect event generator Sometimes this event generator failed to detect a change in the flag status The code relied on the event to reset the flag therefore when an event was not detected the system was left hanging thus preventing any data transmission from occurring The cause of this problem was not investigated but worked around by polling the flag The second algorithm tested polled the flag at one second intervals While this method reliably generated the event it often lost an array of polar co ordinates Investigation of this problem revealed how Citect updates its variable data The Citect variable tags are updated at intervals determined by an algorithm hard coded in the Citect system Therefore when the value of a variable is requested in the visual basic code the value is taken from the local RAM and not over the network from the PLC or embedded PC The consequenc
53. ing two PLC s This dumping operation is monitored by an operator in the control room The control room has a CCTV network to watch the operations An operator will also monitor the operation via a SCADA HMI The existing SCADA system was developed using Citect BEC Engineering 2008b The existing Citect system window gives the operator real time access to information from the PLC regarding the unloading operation Part of this project is to integrate this laser scanning system into the existing site SCADA HMI system allowing the operator to view the information gained from the laser scanner Citect is a fully Integrated HMI Human Machine Interface and SCADA Supervisory control and data acquisition package Citect is used to collect data and provide an interface to control specific equipment such as PLC S or remote terminal units Citect 2008 4 1 SCADA Communications To enable transfer of results and scan data from the embedded PC to the Citect SCADA system a method of data communications had to be established Citect has the ability to communicate with PLC s and other equipment using a large number of protocols The existing sites have Ethernet communications networks established and therefore Ethernet was used to integrate this project into the existing infrastructure MODBUS TCP IP was selected as the application layer protocol The existing sites have PLC s that use the MODBUS protocol MODBUS is also one of the protocols prefe
54. l development using the test model platform triggering was achieved by using a search algorithm Once a set of Cartesian values are calculated the results are then examined to determine if an ore car is present Several simple algorithms were tested for the identification of an ore car The first algorithm compared the mean Y value with a set point but was prone to false triggering The second algorithm searched for the outsides of the ore car by looking for values of y that were the equal the height of the car within a defined range of X values This algorithm improved on the first but was triggered by the ends of the ore car The final algorithm also searched for the inside of the car wall If the search tool found a left side and right side and a left inside and a right inside then the integration would proceed The final algorithm was still prone to false triggering but was sufficient for workshop testing 3 4 Finding the profile Once the Cartesian co ordinates corresponding to the ore car have been ascertained the next step is to integrate Selecting a suitable point to begin and end the integration on the cross sectional profile has also been researched Different strategies have had a significant impact on the accuracy of the final answer 26 In this project one of the key aims is to be able to determine the degree of imbalance between each half of the car To determine this we required separate values for the volume in each side To ca
55. larm flag registers is set Set points values are inserted directly from or to the corresponding Citect tag The density value used to convert the volume of ore to a mass value can also be changed by the operator Density is treated as a set point and is also inserted directly into the Citect tag The ordinal number of the present car extracted from the PLC is also displayed on the Citect HMI as well as the car class type The Citect HMI screen can be seen in figure 4 1 44 Fe Ore Car Laser Scanner 8 03 55 PM 11 5 2008 HIGH_TONNAGE_ALARM High Tonnage Alarm ORE CAR LASER SCANNER CAR No 10 DENSITY _ 1 Ore Car Laser Scanner start 3 9 GB Citect Project Editor Ca w Citect Graphics Builder Sei TE Carscan_080916_1405 YW untitled Paint Figure 4 1 Citect SCADA HMI screen 4 3 Buffering scan data Event timing was by far the most difficult problem associated with data communications over the SCADA network for transmission of charting data Citect is essentially a GUI package to interface to control systems for the process control industry Citect was not really designed to load and store data One of the challenges during this project was to find a way of buffering the data from the laser scanner to be displayed on the Citect GUI MODBUS TCP IP is a master slave protocol and Citect only has the ability to act as a MODBUS master This prevents the laser system from generat
56. lculate the two values for volume integration needed to be performed over each side of the car Firstly integration can be made from a side wall to the centre point and secondly integration is made from the centre to the other side wall The first integration algorithm trialed calculated the integral between fixed points in space Each car half was integrated from the centre of the measurement array at zero degrees to a point in the array which was just outside the ore car This algorithm produced excessively large variations in results The cardboard models had sideways free play between the railways which most likely caused a variation in results Integration needs to follow the ore cars in order to integrate between fixed points on the car instead of fixed points in space An algorithm had to be developed to find a reference point on the cars which would be found consistently in the same place in order to produce consistent results A separate program was developed to simultaneously trial five different methods of searching for a point which could be accurately reproduced This program started from the outer most scan result and searched through the data looking for the side wall of the car The five methods were 1 the first Y value to exceed its set point 2 the first delta Y value to exceed its set point 3 the first derivative of Y to exceed a set point 4 the first delta Y value to exceed a negative set point 5 the first derivative o
57. lted to a frame supported by the office partitioning as shown below in figure 1 4 Ore car models were scaled to one fourth of the size of an actual car based upon mechanical drawings These models were constructed from cardboard boxes The models were given a similar cross sectional profile as shown in figure 1 5 The height of the scanner above the models was also one fourth of the distance expected on a typical installation Not having wheels the models were kept in line using two lengths of aluminum angle 10 Figure 1 4 Laser scanner workshop testing PO te Figure 1 5 Cardboard scale model 11 Although a simple model setup it provided an effective test platform In particular using the cardboard models allowed the testing of all components in the system The cardboard models could be filled with objects of known volume to assist in the calibration of the system and provide test data These models were dragged underneath the scanner using string to represent the cars being ejected from the tipping cell and moving beneath the laser scanner 1 6 Company Profile BEC engineering is an electrical engineering consultancy company BEC is based in Perth with offices in Brisbane and Tanzania The company has approximately 130 staff consisting of six division managers and several other support staff in drafting administration and IT support The management structure is illustrated in figure 1 6 BEC n d The power generation te
58. ly for schematics and switchboard layouts The role of the intern was predominantly detailed software engineering for the embedded PC and Citect SCADA research into effective algorithms interfacing the equipment providing technical information to the installation contractors liaising with the client assisting with documentation and providing feedback to senior staff 13 Figure 1 7 BEC Engineering laser scanner project Team 14 2 0 Laser Scanner The laser scanner uses the time of flight principal to measure an array of distance measurements through a radial arc The measurements are then transmitted over serial communications to the embedded PC Communications connections are made using a 500k baud rate which is not one of the standard Microsoft baud rates and requires a dedicated PCI card For transmission over the serial communications the distance data is encased into telegrams A set of measured values is transmitted after a request is made by the remote devices Requests are made using a command which is also encased into a telegram Figure 2 1 SICK laser scanner LMS 221 2 1 Theory of operation The laser scanner chosen for this project has a fixed laser beam which is reflected by a spinning mirror through a radial arc SICK 2006b The unit can be used to measure distances to points through this arc at fixed intervals of angle The arc size and intervals of measurement depend on the specific model used The Sick LMS 2
59. m at the rail workshop on real ore cars but under a simulated situation Approval was given for the installation of one car dumper as a prototype and then subsequently at four other sites The project aim is to develop the system to comply with the documented functional specifications as well as being a system which is safe reliable and accurate Although the accuracy of the volume and level measurements are sufficient the alignment measurement requires further development The software components for this project also require further testing to ensure compliance with IEEE software standards The longer term goal for the installations is to use the information gained from the laser scanners as feedback to the control system Should the scanner find an ore car which contains excessive residual ore then the control system could reverse the ejection and re tip the ore car immediately Having spent a great deal of time researching the limitations of these laser scanners and establishing reliable communications with the scanners the company now has a template which can be used to implement the scanner for other uses These scanners are commonly used for anti collision systems robotic applications and profiling the contents of conveyor belts SICK 2006b Another important part of this project was the establishment of communications between the embedded PC the PLC and SCADA systems While these were achieved with the common MODBUS TCP IP protocol it st
60. o has the ability to be able to access Windows Active X objects The charting component of this system was initially developed using examples of source code from Gigasoft 2008 implemented into visual basic 6 because there were no code examples for citect VBA or Cicode Once a working algorithm had been developed the source code was adapted into Citect VBA The Citect SCADA in this project has three Citect VBA functions developed for controlling the graphing package The first one to initialize the chart the second function is to add new chart data whilst the third will refresh the chart for a new ore car There is also a main function from where the charting functions are called The three function are now discussed in detail The initialize function is called when the window is opened by the user This function sets all the necessary variables and invokes all Active X methods required to draw the chart on the window A number of plot styles are available on the package In this project a polygon plot style is used which inserts a coloured polygon between each plot point The polygon colour is determined by the Y value therefore the chart is presented as series of contours A minimum value has been set to 500 millimetres preventing any values below 500 from being displayed on the chart Thus the ground profile is prevented from being displayed All scales are set and fixed The axis markers and legend have also been turned off The number of X an
61. of the chart methods which automatically shuffles all the previous values along one point in the Z plane The oldest values are deleted and the latest set added This function also redraws the screen The refresh function replaces all the y values with zero before redrawing the screen This gives a blank screen when a new ore car is under the scanner This function is called whenever a new car flag is set by the embedded PC on the MODBUS slave register This flag is reset within Citect Function NextVal is invoked by the Citect event generator every second whenever the Car Scanner page is open This function checks the new data flag for new data and then proceeds to unload the MODBUS slave registers containing the arrays of polar data Once the array of polar co ordinates are converted to Cartesian co ordinates these values are passed to the update function The Citect SCADA screen also displays the results of the last car that was scanned including the quantity of residual ore imbalance of ore the level of the car and degree of car misalignment These values are displayed directly from the appropriate tags Alarm set points for these values can be entered onto the Citect HMI screen by an operator The embedded PC contains the code to determine if any of the values exceeded its alarm set point Should a set point be exceeded then the appropriate alarm flag will be set by the embedded PC Citect will generate an alarm event when one of the a
62. oint in space to determine if the car would hit the tipping cell when the car was pulled back in This algorithm was repeated over the last quarter of the car The results had to be taken as an average of multiple scans in order to reduce the effect of instrument noise By comparing the value with a fixed point the error would be immediately halved The output value for this algorithm is then the difference in centimetres in X value between the centre on the car and the centre of the rails This value represents the points one eight from each end of the car 33 Figure 3 8 Car Alignment 3 6 2 Level Another algorithm which was considered to help determine if a car is correctly aligned is to compare the height of the side walls This algorithm searches through all scan results for any y values above the set point By taking a mean of all those on one side of the car and comparing this value with the mean of the second side of the car a difference in height is found This difference represents degree of tilt and can be used to help determine if the car is on the rails or if the car is not sitting on the bogie correctly This algorithm is reasonably robust but can t distinguish between the car and the ore in the event that the ore in the car is higher than the side wall of the car In the acceptance testing this algorithm will be tested by chocking up one side of the car The car body is not permanently attached to the bogies and relies only
63. on in the car scanning software to change configuration settings In the event that the scanner will require replacement a procedure for changing the mode and baud rate is detailed in appendix ii 2 2 Data Communications The SICK LMS 221 unit can communicate using either RS422 or RS232 serial protocols RS232 is limited to 10 metres while RS422 can go as far as 1200 metres SICK 2006b In all the installations the distance from the laser scanner to the equipment room is greater than 10 metres and therefore R8422 communications will be used RS 422 transfers data using two balanced pairs which has better tolerance to noise allowing greater cable distances and faster data transmission Lammert Bies 2008 SICK 2006b also specifies that cables must be screened twisted pair To meet this specification a category 5 foil screened twisted pair cable has been selected and installed 16 2 3 Baud Rates The SICK laser scanner has a choice of four baud rates 9600 19200 38400 or 500 000 The selected baud rate must be sufficiently high enough for all the data from a scan to be transmitted between the requests for scans Should a request for values be made before the completion of a previous scan then the transmission is aborted resulting in an invalid telegram and 100 loss of data Minimum scan intervals were tested for each baud rate with the results shown below in table2 7 Baud Rate Minimum scan Interval ms 9600 1000 19200
64. on the extrapolated result The third algorithm tested was a combination of the two previous algorithms This algorithm searched through the data to find the furthest most point from the centre that exceeded the set point The mean of this value and the two adjacent inner values was used to calculate the reference value for X This method was found to be more immune to changes in the height of the side wall and furthermore able to produce reliable results where contents of the car were higher than the side wall While the final result did not determine the X co ordinate as accurately as the previous algorithm it was selected as the most suitable candidate to make the system as robust as possible 3 5 Profile Integration Two integrations are performed to convert the data arrays to volume The first integration finds the cross sectional area of the ore car for each scan This integration is performed numerically using the trapezoidal rule This rule has a good level of accuracy well beyond the requirements of this application The trapezium rule is also easy to convert to a form that can be used where the AX values are not equal Because the co ordinates are converted from an array of polar values the AX between points has a substantial variation This variation is too large to ignore According to Croft 2001 the integral is the sum of individual areas of trapezia where Ax AreaOfTrapezia gt nn tya Easily adapted for changing AX values
65. operations staff and to facilitate a reduction of economic loss The laser scanner is a non contact measurement device It contains a fixed laser and a rotating mirror to measure distances in a radial arc The time of flight principal to determine the distance based from the return time of a pulse of light Arrays of polar co ordinates are outputted from the scanner to the embedded PC which converts these to sets of Cartesian co ordinates before integrating these to determine volume The volume of residual ore is the difference between the scanned volume of a given car with an empty car Volumes are calculated for either side of an ore car and added to give the total volume of residual ore The individual volumes are compared to determine the degree of imbalance with a measured car Both these values are scaled by a density value obtained from the SCADA system before they are displayed on Citect HMI and recorded into a database Other measurements taken when using the laser scanner are the degree of level and the alignment of the cars Level is measured by comparing the heights of the two side walls Alignment is measured by comparing the horizontal centre of the cars at both ends with the centre of the rails These measurements are taken to assist in finding cars which may be damaged or derailed A significant portion of this project included the establishment of communications between all the hardware components Communications with the scanner w
66. rammed to accommodate the laser scanning system BEC 2008 b An interface will also need to be developed for communications between the SCADA system and the embedded PC A block diagram of the components for this system is shown in figure 1 2 Laser Bees SCADA Scanner Ethernet r EEA PLC Embedded PC z Figure 1 2 Hardware and communications block diagram This report describes the development and interconnection of the various components which had to be researched procured adapted or designed to generate an overall working solution Specifically this includes 1 Interfacing the laser scanner with the embedded PC 2 Interfacing the existing PLC with the embedded PC 3 Interfacing the existing Citect SCADA with the embedded PC 4 Development of software for the embedded PC 5 Building a Database on the embedded PC 6 Modification to the existing Citect software The scope of this report covers the research and development of a stand alone turnkey product While this system has been developed for a specific client this report is of a general nature intended only to cover the research and development of a product to deal with a common problem in the mining industry At the time this report was compiled the system had only been tested in a workshop environment and had not been tested on a working mining operation 1 4 Concept A laser scanner is a non contact distance measurement device which w
67. ring services in the utilities sector including power generation utilities independent power producers customer owned generation co generation embedded generation and renewable energy companies BEC also has ongoing involvement with other utility services for water transport and treatment plants gas transmission pipelines compressor and delivery stations BEC n d 12 Managing Director Plant Integrity Manager Automation Manager Process Control Manager Power Distribution Manager Power Generation Manager Operations Support Manager Figure 1 6 BEC Engineering Senior Management 1 7 Project Team Members The key staff from BEC involved in the project addressed by this report can be seen in figure 1 7 Overall the project is managed by the Automation manager who developed the initial proposal for the client developed the scope of work functional specifications tender documents procured the necessary hardware and software components The lead engineer is the principal point of contact for the client manages installation contractors and scrutinizes detail design supervises technical and engineering staff The responsibility for cost control quality management and document management fell upon the project manager A senior electrical engineer provided detailed design and developed testing procedures The draftsperson involved in the project provided detailed drawings specifical
68. rred by the client MODBUS TCP IP was selected in order to keep the number of protocols used on site to a minimum OPC was also considered but rejected by the client MODBUS is a master slave protocol and Citect only has the ability to act as the master Normally a PLC is the slave Therefore the embedded PC will need to act as the MODBUS slave to communicate with Citect The relationship between application layer protocols are shown in figure 3 10 41 For the embedded PC to communicate using MODBUS TCP IP protocol as a slave a commercial software package was also used A number of suitable software components were assessed to find one suitable for this project The selected component was the MODBUS TCP slave also from Automated solutions Automated Solutions 2008 This product is of industrial quality and is included into the embedded PC software as an Active X object The object emulates a MODBUS slave device such as a PLC or RTU The object contains MODBUS standard register types such as input registers holding registers inputs and coils The Active X object accepts MODBUS standard read and write instructions from a MODBUS master device This product is also the complement to the MODBUS master software used to communicate with PLC s discussed in section 3 9 One development licence was procured for use in this project Each variable used in Citect needs to be defined setup and configured Citect documentation refers to variables a
69. rrent alarms No of requested alarm Residual Ore Tonnage Tonnage Difference Car Level Car Skew Density used for calculations Residual Ore Tonnage alarm se Tonnage Difference alarm set D Car Level alarm set point Car Skew alarm set point 64 40101 string CARSCAN_CARNUM_V Car Number 40151 string CARSCAN_CARTYPE_V Car Type Integer 40101 40220 array CARSCAN_POLAR1_V polar coordinates of scan results Integer 40301 40420 array CARSCAN_POLAR2_V buffer 1 Integer 40501 40620 array CARSCAN_POLAR3_ V buffer 2 Integer 40701 40820 array CARSCAN_POLAR4_ V buffer 3 Integer 40901 41020 array CARSCAN_POLAR5_V buffer 4 Integer 41101 41220 array CARSCAN_POLAR6_V buffer 5 65 Appendix x sample of raw data test Citect communications This data is a sample taken from a text file which was used to help understand and debug the way in which Citect retrieves data from a slave device using MODBUS Information from the MODBUS event function was simply saved into the file Apart from the time the lines show the size of the buffer at the time the instruction and the registers affected 10 23 2008 2 57 31 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 10 23 2008 2 57 32 PM 1
70. rt representing 50 degrees to 50 degrees in the two dimensional polar plane The following equations give an array of Cartesian X values and an array of Cartesian Y values from the array of polar co ordinates X Polar Si MT 360 Y Polar Coj ST 360 Where Xn is the n th Cartesian X value in the array Yn is the n th Cartesian Y value in the array Polarn is the n th Polar length value in the array nis the array index 25 Once a set of Cartesian values are determined the next step is to invert the Y values The co ordinates are initially relative to the scanner therefore to make the points relative to ground and positive the height of the scanner above the ground needs to be factored into the equations The height from the scanner base to the ballast is typically taken at the base of the railway line as shown in figure 3 2 This height value is read from the text configuration file at start up The individual Cartesian Y values are subtracted from the height value to give the Cartesian values relative to ground level instead of relative to the scanner 3 3 Triggering To be able to determine an accurate measurement for the volume of residual ore the PC must know when to integrate the scan data Integration needs to be performed only when an ore car is under the scanner and the car is moving At the site installation information obtained from the PLC registers can be used to determine a suitable start and finish time For the initia
71. s S The volume of residual ore is the difference between integral of the scanned ore car with the integral of an empty ore car These values are stored in the database for each type of ore car and retrieved with the ore car length when required Integral values for empty ore cars will be determined by statistical information obtained from the results database once the system is installed on the car dumpers Separate values of volume are calculated for each side of the ore car The sum of the two volumes gives us the total volume which is multiplied by the density to give a total mass of the residual ore The value of density is obtained from the set point database on the embedded PC This value can be altered by operations staff using SCADA HMI An absolute value of the difference between the volumes on either side multiplied by the density gives us the tonnage imbalance 32 3 6 Alignment The eventual aim for this equipment is to use the quantity of residual ore to invoke a procedure to re tip an ore car A re tip decision should only be made on condition that the car is still on the rails and has not been damaged during the tipping process There has been an incident where the bogies of the rail ore car had been completely broken away from the car during the tipping process If a car has derailed or a bogie damaged when it is pulled back into the tipping cell there will be insufficient clearance between the cell arms which hold the ore car an
72. s tags Each tag needs to be configured in a database so that citect knows how and where to retrieve the information Normally when communicating with a PLC a tag would represent a system variable such as a process variable manipulated variable etc Tags can be almost any of the basic types such as integers real numbers booleans or strings Each tag is defined in terms of its name variable type IO device protocol and MODBUS slave register number Appendix ix has a full list of tags with their variable type MODBUS register number and a description of their purpose To keep network traffic to a minimum raw polar co ordinates are sent from the embedded PC rather than the Cartesian co ordinates which occupies more bytes These co ordinates are converted in to Cartesian values in Citect before being sent to the charting package In the polar array each co ordinate measured is separated by half of a degree so the polar angle is implied by the position of the co ordinate in the polar array This means that polar co ordinates only require one array of values where as Cartesian co ordinates required two arrays for X and Y values The polar co ordinates are also of type integer whereas Cartesian co ordinates are of type double Overall a single Cartesian co ordinate will require 8 bytes of data and polar co ordinates will only require 2 bytes Conversion to Cartesian co ordinates is calculated by Citect for use in the Citect GUI in addition to the conver
73. sion done in the embedded PC for the calculations 4 2 Charting The main feature of the graphical user interface is the three dimensional chart of the cars This chart will show a profile of the cars on the X and Y axis while the third Z axis will be time The plots will be taken while the ore car is moving so the Z axis will resemble the third dimension of the ore cars 42 The chart has been developed using the graphing package proessentials V6 from Gigasoft Gigasoft 2008 A number of commercial charting packages were assessed to find the most suitable one for incorporation into this system Proessentials was developed for use by the science and engineering industry Gigasoft 2008 Proessentials has a number of different chart and graph types to select from Gigasoft 2008 Implementation into the system software is via an Active X object Gigasoft 2008 It was selected because of its ability to do three dimensional charting with real time updates One development licence was procured as part of the normal process for equipment procurement Part of the package from Gigasoft includes a number of code samples in several languages including several C variants and visual basic 6 This project uses the 3D real time charting component of proessentials and is programmed using the visual basic development kit within Citect Citect is programmable using its own native language Cicode or a variant of visual basic called Citect VBA Citect als
74. ssessment of the amount of ore which can be seen in the car after being pushed under the scanner Test for level Using hydraulic bottle jacks or equivalent equipment one side of the ore car body will be jacked up from the bogie Chocks will be placed between the bogie and the body so that the side wall is 100mm higher than normal Confirm that the measurement for level is approximately 100 mm after the car has been pushed under the scanner 69 Test for front alignment A length of timber at least 2 meters long and 100mm thick is to be clamped to the top of the side wall on the outside After the car has been pushed under the scanner confirm that the measured value for alignment is approximately half of the thickness of the wood Repeat above for each corner 4 of 70
75. tained in table 2 11 The send telegram is always 8 bytes long while a receive telegram can be up to 518 bytes long SICK2006a Referring to table 2 10 and table 2 11 the frame components of the telegram are the same in both the send and receive telegrams The start of text is the value 02h which is a preamble to identify the start of the transmission The address byte is the bussable address which is not used on this model and is 00h by default The length values are the number of subsequent data bytes not including the check sum Byte 5 is the command response byte with the associated data placed into byte 6 The checksum is used for error detection for which the algorithm is detailed in SICK 2006a In this project the scanner has been pre configured using the SICK LMS 400 user software SICK 2006c Therefore the embedded PC will generally only require commands for the normal day to day operation but not configuration commands Specifically these commands will be to request scan values from the scanner and to turn the laser on and off SICK 2006a contains a detailed discussion of a large number of telegrams which can be used to communicate with the laser scanner Frame Commands amp Data Frame Byte 1 2 3 amp 4 5 6 788 Position Description Startof Address length Command Data Checksum text Table 2 10 Command Telegram Structure SICK 2006a Frame Response amp Data Frame Byte 1 2 3 amp
76. tandard serial port UART is driven by a crystal controlled clock speed of 14 7456 mhz The 500k serial port developed for the SICK unit has a clock speed crystal frequency of 16 000 mhz The slightly higher frequency for the clocking frequency will now give a true data rate of 500 000 baud With the combination of using a 16 MHZ clock with the times 8 jumper closed and the MSCOMM32 baud rate set to 56 000 allows the PC to communicate with the laser scanner at 500 000 baud A table of baud rates for the high speed communications card is shown in table 2 9 SICK 2002 Target baudrate gt 9600 19200 38400 500k MOXA and Quatech 1075 2212 4300 56000 Use baudrate card PCMCIA 2150 4301 8602 115000 SIO RS485 SICK Table 2 9 Baud rate conversion table SICK 2002 18 2 4 Retrieving scan values The data format used in communications from the laser scanner takes the form of telegrams Each byte has 10 bits with 1 start bit 8 data bits and 1 stop bit The bytes are assembled as blocks Control command blocks are sent to the scanner or response blocks are received from the scanner These blocks of data sent to the scanner are referred to as send telegrams while the blocks received back from the scanner are receive telegrams During normal operation the laser scanner will send a response telegram in reply to a command telegram The structure of a send telegram is contained in table 2 10 The structure of a receive telegram is con
77. ter of finding the value half way between the two side walls Y co ordinates for the two sides X values and the centre value are found from linear extrapolation of the neighbouring values W SL FLLS Figure 3 7 finding the side wall The second algorithm which was developed used linear extrapolation to find the outside of the car This method searched through the scan data to find where the profile Y co ordinates crossed the set point on the outside of the ore car Linear extrapolation was used to find the X value that would correspond to the set point from the two adjacent points from either side of the set point This X value could now be averaged over the first quarter of the car reducing the effect of instrument noise The calculation is repeated for the second side 30 These two algorithms were compared by scanning a stationary object while watching for variation in results The raw test data can be seen in appendix v This illustrates that the second algorithm has far less variation in final value with less than 1 millimetre while the first algorithm has approximately 3 millimetres variation It was also noted that the first algorithm exhibited incorrect values when the contents of the car exceeded the set point height The second algorithm was vulnerable to changes in height of the side wall Because the X value is extrapolated from known Y co ordinates a change of Y value has a significant impact
78. the upright position the residual ore became dislodged and fell to one side of the car These cars have a tare of 18 tonnes which can hold up to 130 tonnes of ore When the ore has been piled up on one side of the car the shift in centre of gravity can be sufficient to cause the car to become derailed BEC 2008a Train derailments have a serious economic impact on business operations Apart from the damage to railway infrastructure and rolling stock derailments also cause loss to production and can pose series threat to staff safety This project discusses the development of a system to determine if ore has been left in a car after it has been emptied This system will help to prevent train derailments leading to a safer working environment and greater production levels BEC 2008a 1 2 Objective The two main objectives of this project are staff safety and economic loss prevention The proposed system uses a laser scanner to profile the cars as they are ejected from the tipping cell This equipment will be used in the early detection of a situation which could lead to a train derailment In the event of excess residual ore or where the ore is piled on one side of the ore car an alarm event will be generated to alert operations staff of the situation Operations staff will then have a numerical measurement for the amount of residual ore and a 3 dimensional chart showing the profile of the ore car contents This will enable staff to make an informe
79. tomated Solutions 2008 This product is an Active X component which emulates a MODBUS master over a TCP IP network A block diagram of communications protocols is shown below in figure 3 10 While the protocols are MODBUS TCP over Ethernet the embedded PC acts as the master for communications with the PLC and the slave for communications over SCADA SCADA communications is discussed in section 4 1 39 Embedded MODBUS TCP Master MODBUS PC Laser OFF TCR Slave Scanner Ethernet MODBUS TCP Master A MODBUS TCP Slave Figure 3 10 Application Layer Protocols 3 10 Error Handling The embedded PC software includes an event log Any detectable abnormal events significant events or run time errors are recorded into a text based events log file Normal events which are recorded include start up and shut down processes Losses of communications with any of the laser scanner SCADA PLC or the database are also recorded Database anomalies such as missing corrupt or invalid data are also logged Despite extensive testing in the event that a visual basic error does occur each non critical function has an error handler to allow the program to continue operating The error handler will record the function where the error occurred and the error message Appendix viii contains a sample taken from the event log while the software was being tested 40 4 0 SCADA The car dumper is highly automated and controlled directly by us
80. which caused the effluent to run down an embankment into the electrical cable pits Contamination of the pits created a hazard for the staff contracted to install the equipment Therefore installation was delayed until the plumbing problem could be rectified There was also a delay concerning the location of the laser scanner The laser scanner had to be mounted in a position over the cars as they are ejected from the dumper The dumping operation occurs continuously throughout the day and cannot be stopped without economic loss The installation of the scanner had to be performed using an elevated work platform in order to provide a safe working environment to the installation contractors Time taken to organize this equipment resulted in further delays to the project A further subsequent delay was at the request of the client As the installation progressed and an increasing number of client management became aware of the project the client requested a hold on the project in order for the specifications to be reviewed again The installation delays had an impact on the timing of the detailed design work Some of the staff who was initially allocated to this project had dispersed to other projects Installation delays also meant that testing had to be performed using the cardboard models instead of on site testing The result was that the software was at a reasonably completed stage of development before the system could be tested on site 49
81. y open rolling stock from a mining operation to a processing plant or port for export This project focuses on the unloading operation of these railway ore cars The cars are unloaded by rotating the entire car allowing the contents to fall into a hopper The ore is then transported to a stockpile by conveyor belt During this operation cars are disconnected from the train pushed into the cell and then rotated upside down Once the contents have fallen out the cars are returned to the upright position and ejected from the cell Figure 1 1 shows an example of an unloading cell This operation is fully automated using a PLC but monitored from an operations room via a CCTV network and a SCADA network Figure 1 1 Ore car unloading cell When ore becomes damp it turns to mud and can cling to the inside of the car This ore does not always completely fall out when the car is inverted During normal operation if excessive amounts of residual ore are still present in the car after unloading the car can be returned to the tipping cell and re emptied A decision to re tip an ore car lies at the discretion of operation staff who are watching the unloading process via CCTV closed circuit television The re tip operation is invoked manually by staff if necessary BEC 2008a Failure to detect cars with excessive amounts of residual ore has resulted in several serious train derailments Investigations have found that as the ore car was being returned to

Download Pdf Manuals

Related Search

Related Contents

Bodenreparaturset für Fliesen Bedienungsanleitung – Seite 1  DA-8100    User Manual iView  Benutzerhandbuch  231 lankofuge surface  2 Menu Wii - Nintendo of Europe  KitchenAid 4KPTT890 Toaster User Manual  fulltext01  MANUAL DE USUARIO  

Copyright © All rights reserved. Failed to retrieve file