37361131_APC_AddOns_en
Contents
1. Version V1 0 July 9 2009 39 48 Copyright Siemens AG 2009 All rights reserved SI E M E NS Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 ModPreCon APC AddOn Products 37361131 4 3 Illustrations for ModPreCon Figure 4 8 System architecture for DCS embedded predictive controller ModPreCon pe Model Predictive Controller MPConsim TIC811822 i Mse2l 7o 12 37 E P4 Track Onf 4 Track Onf OS Clients E trakon ModPreCon OS faceplate gt operation amp control Toniroer make ee 2 n 251000 un Dei Controle Deren 1 MPC Configurator Engineering gt modelling amp Station ES controller design generation of user EM data block Ethernet Industrial Ethernet Fast Ethernet OS Server redundant i i OT CE AI E 13 BEE er EN ilm Is TU ele LIL UFER l ModPreCon function block and user data block f gt runtime calculations Version V1 0 July 9 2009 40 48 Copyright Siemens AG 2009 All rights reserved SI E M E N S Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 ModPreCon APC AddOn Products 37361131 Figure 4 9 Model predictive controller ModPreCon as PCS 7 function block TIC811822 Controlled Variables Disturbance Variable Setpoint Filter Manipulated Variable Track 0 MYlTrkOn Track Of oetpoints Manual MVs Pe nee2. B F Gon Oo 50 E T1 1 Actual Value pee EA 2 C 79 6343 ER t Manipul E MU Hide 3 f 20 5858 50 MU Lo E3 Manti ManLoOx 100 0 Conzene MS Rel an Kn EN aT 7 49 30 7 50 00 7 50 30 Zen 3P Exeri anas ose mes sv l zs Es SP Exti Proce ECCO New Archive Open Archive Close Archive DETTE DRE E SP Ext SP InHi EH SP InL PU Ox ER I PU Unit x MU Unit Se P t Controll t i rocess parameters ontroller parameters m VZ2 model damping 1 400 PID PI P Een 3 407 Proportional gain 3422 0 687 0 282 ES mE Echec Integration time 7 528 3 884 ga Model fit 59 930 BE Derivative time 2 085 s EH ee Time lag 1 466 s CPI m Recovery time 21 691 s B A kaa e e C IE Max sampling time 0542 s ES Please select the desire controller type Se PRE T E e and press Next mx lt Zur ck Weiter gt Abbrechen Hilfe 8 A EEE iM Press F1 for help A Sheet1 MERON Version V1 0 July 9 2009 15 48 Copyright Siemens AG 2009 All rights reserved SI E M E N S Optimization of PID controllers RaPID by Ipcos versus SIMATIC PCS 7 PID Tuner APC AddOn Products 37361131 2 4 Hints for Selection of Appropriate Product 2 4 1 Arguments for Application of PID Tuner Seamless integration in PCS 7 No software license costs Lower engineering costs If the requirements for the application of PCS 7 PID Tuners are fulfilled and you are satisfied with the tuning results you don t need RaPID
3. Copyright Siemens AG 2009 All rights reserved SI E M E N S Warranty Liability and Support APC AddOn Products 37361131 Note The application examples are not binding and do not claim to be com plete regarding the circuits shown equipment and possibilities The soft ware samples do not represent a customer specific solution They only serve as a support for typical applications You are responsible for ensur ing that the described products are used correctly These application ex amples do not release you from your own responsibility regarding profes sional usage installation operation and maintenance of the plant When using these application examples you acknowledge that Siemens cannot be made liable for any damage claims beyond the scope described in the liability clause We reserve the right to make changes to these application examples at any time without prior notice If there are any deviations be tween the recommendations provided in these application examples and other Siemens publications e g catalogs then the contents of the o ther documents have priority Warranty Liability and Support We accept no liability for information contained in this document Any claims against us based on whatever legal reason resulting from the use of the examples information programs engineering and perform ance data etc described in this application example shall be excluded Such an exclusion shall not apply in the
4. Product Information INCA PID Tuner alias Ra SIMATIC PCS 7 PID Tuner PID Robust Advanced PID Control Software producer IPCOS NV Leuven Belgium Siemens AG I IA AS and Boxtel Netherlands http www ipcos be Form of delivery External product in add on Since V7 0 integral part of catalogue PCS 7 toolset before option package with extra charge Table 2 2 System architecture INCA PID Tuner alias Ra SIMATIC PCS 7 PID Tuner PID Robust Advanced PID Control Integration in PCS 7 Separate software tool on Integral part of PCS 7 ES external PC Table 2 3 Usability INCA PID Tuner alias Ra SIMATIC PCS 7 PID Tuner PID Robust Advanced PID Control Call Windows start menu Via context menu in CFC of PID controller Coordination of tuning tool No support by tool Tick mark Enable Opti and plant operator Process excitation is mization in PID face operated manually in plate faceplate of controller or During process excita excitation signals are tion the PID block is read from a file remote controlled by tuner software User guidance Interactive Windows program Software assistant wizard with numerous menus and with pre specified sequence numerous user specified of steps parameters Number of parameters to be specified by user is mini mized Version V1 0 July 9 2009 9 48 Copyright Siemens AG 2009 All rights reserved SIEMENS APC AddOn Products Table
5. plants that are already oscillating without feedback control or show non minimum phase behavior i e after a manipulated variable command they start running in the opposed direction first e You require especially high control performance and are therefore prepared to spend a lot of time for the fine tuning of individual con trol loops As a tool from experts for experts RaPID offers a lot of features functions and tuning parameters July 9 2009 16 48 Copyright Siemens AG 2009 All rights reserved SI EMEN S Soft Sensors Based on Artificial Neural Networks Presto by Ipcos versus SIMATIC NeuroSystems APC AddOn Products 37361131 3 Soft Sensors Based on Artificial Neural Networks Presto by Ipcos versus SIMATIC NeuroSystems Many process engineering plants suffer from the fact that for important quality parameters of intermediate or end products there are currently no low cost low maintenance reliable and fast sensors available on the mar ket The application of online analyzers or the execution of laboratory analyses is expensive and even worse it takes time so that it is typically too late for efficient control actions to achieve the desired specifications The application of model based estimation methods is an alternative solu tion in such cases because they make use of process values that can be directly and easily measured in order to predict quality parameters This requires the existence of an appropria
6. 7 23 2003 4 57 07 PM Bias 7 0480E 01 Lab Time 1772372003 4 52 07 PM TT Lab sample 14 8331E 00 TJ405 ER PV 118 90 0 TJ406_ER PY 118 55 0 55 H HAC BIAS PV 0 70 55 RX HAC NEUR P 3 18 AIT91 EF PV 4 83 0 55 HX HAC pv 3 88 0 bad NN output 0 00 no bias updates 0 00 NN status 0 00 Version V1 0 July 9 2009 24 48 Copyright Siemens AG 2009 All rights reserved SI M N S Soft Sensors Based on Artificial Neural Networks Presto by Ipcos versus SIMATIC NeuroSystems APC AddOn Products 37361131 3 3 Illustrations for NeuroSystems Figure 3 4 System architecture for SIMATIC NeuroSystems A ATE TEN piel pies AN Dasa Parisa E OS Clients dei mew mem mj pe P 80 v NeuroSystems tool Engineering NeuroSystems faceplate gt network training Station ES gt operation amp control creation of user data block Ethernet Industrial Ethernet Fast Ethernet OS Server redundant us Zu mi 1733 HII 11 Kuhn OP LPO eee Poe E T oL UP ER EO a mam hin T ma cmi NeuroSystems function 4 block and user data block LE LE um HE EL mr n x Version V1 0 July 9 2009 25 48 SI E M E N S Soft Sensors Based on Artificial Neural Networks Presto by Ipcos versus SIMATIC NeuroSystems APC AddOn Products 37361131 Figure 3 5 Function block NEURO_64K in SIMATIC CFC i CFC CFC 1 FM Project SIMA TIC 400 1 4CPU 41
7. I IA AS and Boxtel Netherlands http www ipcos be Delivery form External product in add on Since V7 0 1 integral part of catalogue is typically sold in PCS 7 toolset as part of conjunction with engineering APC Library respectively services as turnkey solu Advanced Process Library tion Table 4 2 System architecture INCA MPC Ipcos Novel SIMATIC PCS 7 MPC bzw Controller Architecture ModPreCon Integration in PCS 7 Separate software tool on PCS 7 function block with external PC faceplate and configuration tool Runtime algorithm INCAEngine as OPC DA PCS 7 function block MPC or client on Windows PC with ModPreCon connection to Operator Sta The function block requires tion requires an Ipocs Data considerable computing Server and Scheduler as power and a separate user Version V1 0 July 9 2009 30 48 Copyright Siemens AG 2009 All rights reserved SIEMENS Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 ModPreCon APC AddOn Products 37361131 INCA MPC Ipcos Novel SIMATIC PCS 7 MPC bzw Controller Architecture ModPreCon runtime environment data block for parameteriza Dedicated APC interface tion function blocks are required The function block can typi as infrastructure in the AS cally be called in low priority Those are provided by CC cyclic task inside the CG as re usable solution SIMATIC controller Starting from PCS 7 V7 1 all controller blocks
8. MVs than CVs Only an online opti mization can make goal oriented use of these degrees of freedom e g by targeting economically optimal values for the MVs Control problems with control targets of different priority in a fixed rank ing order Only a hierarchical online optimization can make sure that targets of lower rank are considered only if all targets of higher rank are already fulfilled completely Example plant safety has higher rank than product quality product quality has higher rank than reduction of resource consumption Remark the term safety in this context refers to staying within limit values it does not refer to replacing dedicated safety oriented controllers safety shutdowns etc Numerically stiff control problems where inside of a multivariable process very fast and very slow part transfer functions are interacting In these cases dedicated model structures in INCA like e g state space models are helpful In the meanwhile there are two extensions of INCA that are not yet listed in the PCS 7 add on catalogue but could in principle be interfaced to PCS 7 similar to INCA Version V1 0 INCA NL for nonlinear processes like e g batch reactors or cristallers Existing nonlinear physical models are used primarily instead of the ex perimental identification of linear models from learning data INCA MPC4Batch with special features for batch processes Model and controller parameters are adapted t
9. Mode switch EL Version V1 0 July 9 2009 41 48 Copyright Siemens AG 2009 All rights reserved SIEMENS Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 ModPreCon APC AddOn Products 37361131 Figure 4 10 ModPreCon faceplate on PCS 7 Operator Station Ed EL 9 2008 10 17 39 AM 1Sim SimOn600 Model Predictive Control 2x2 Commissioning 1C611622 Manual gt Auto Demo i i 120 ar fe 1481 SP1 gt 150 1 Ic SP2 250 Track Onf SP CY v N mv2 4477 qe Track Onf 7 E M M eon d e 265 1 265 1 265 1 265 1 255 1 265 250 4 250 4 250 4 250 4 250 4 250 n 200 200 4 200 200 1 200 4 200 150 1150 4 150 4 150 1 150 150 100 100 7 100 4 100 100 100 0 0 0 0 0 0 wa 09 1 9 08 10 15 00 AM 10 16 00 AM 10 17 00 AM 5 AS Hale Se AP P T8 E 8 z ae 4 Version V1 0 July 9 2009 42 48 Copyright Siemens AG 2009 All rights reserved SI E M E N S Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 ModPreCon APC AddOn Products 37361131 Figure 4 11 GUI of MPC configurator in second working step display of process model and specification of controller parameters ModPreCon process model ji Ee la xl Raw data file Dy MN benchmark s11 txt Fit 68 Identified system step responses CW importance i as MY move penalty 4 cv Parameters Controller sample tim
10. RaPID is a tool from experts for experts i e RaPID can be success fully applied only by control engineering specialists with the appropriate theoretical background RaPID takes some time to get familiar with the software the manual contains more than 100 pages and is available in English only 2 4 2 Arguments for the Application of RaPID If you are using an older PCS 7 Version lt V7 0 and PID function blocks that are not part of the Standard Library of PCS 7 RaPID is rec ommended because the PID Tuner of older versions is only applicable to standard PID function blocks of the PCS 7 Library In principle any 3 2 party PID tuning tools could be applied The following reasons might justify the purchasing and application of RaPID in special situations although the PCS 7 PID Tuner is also applicable Version V1 0 e You impose very precise requirements how the controller should work in certain situations i e you want to design and optimize the controller for a well defined disturbance scenario or for a well defined setpoint trajectory e g a typical setpoint step from x to y e You impose very special requirements with respect to robustness of the control loop gain and phase margin or the noise sensitivity controller gain at high frequencies RaPID allows for detailed specifications with respect to controller optimization e You are dealing with controlled plants showing extraordinary dy namical behavior e g
11. for one CV of an INCA controller in form of an Excel table KA Microsoft Excel INCA Glass csv i File Edt Waw Insert Format Tools Data Window Help Acrobat DHAa aay sm At oem BIU E 2 as AT a cy 10 Exename MM 01 xj MEIE ar x fe BL i ow 7 Ea t ie Me M Copyright Siemens AG 2009 All rights reserved z5 TOP Cf PV I5 TOP GfPWV 334 1 75 TOP Cf STATUS 15 TOP Cf STATUS E 75 TOP Cf VALLOW 800 SIT 7 TOP Cf ENGLOW 900 7 TOP Cf IDEAL I5 TOP Cideal 25 TOP Cf OPERLOW Z5 TOP C operlow 7 TOP Cf OPERUPP 75 TOP C operupp 75 TOP Cf USE 75 TOP C use 401 T z5 TOP Cf ENGUPP 1020 lana T 75 TOP Cf VALUPP 1200 AT3IT 25 TOP Cf CRITICAL 1 ADALT z5 TOP Cf ZOMERAMK 1 05 T 25 TOP Cf ISSZONEWT 1 ADB T 75 TOP Cf IDEALRANK 3 ADF T 75 TOP Cf IBSIDEALWT 1 T 26 TOP Cf IDYNW T 1 ABIT 75 TOP Cf TRAJECT 1 1 A10 T 75 TOP Cf GAIN 1 1 411 T zb TOP Cf DELAY 1 1 1 412 T i5 TOP Cf RAMPINGON 1 413 25 TOP Cf RAMPLIMLFR I5 TOP _C StorRmp MAJI zb TOP Cf RAMPLIMLOV Z5 TOP C EtoRmp 415 T 5 TOP Cf IDLTRACKINGON 1 AGIT 25 TOP Cf INTERNALIDEAL 1 D 25 TOP Cf IMTERNALIDEAL Always 75 TOP CKIDEAL D 75 TOP Cf POSTSS Always 75 TOP C SSTARG 419 0 ib TOP Cf ACTWE Always 5 TOP C ACTIVE 420 T z5 TOP Cf MODELCYCLE 5 T 25 TOP Cf INTERMITON 1 T zb TOP Cf NOTMEAEON 1 4z3 T F5 TOP Cf NEWPY 1 A204 4 4 FK HAINA Glass e uo Ready Opo pF po dum DE
12. redundant SIMATIC hardware More easy integration in PCS 7 Less software and engineering costs For Presto the integration into the DCS and model engineering require an amount of effort similar to the application of an external model pre dictive controller e g INCA in section 4 In general the Ipcos tool is a tool from experts for experts similar to RaPID and requires the appropriate time to get familiar with the soft ware and the theory behind while the less cost intensive Siemens tool has advantages with respect to usability For smaller soft sensor applications static models with 2 5 input vari ables SIMATIC NeuroSystems is completely sufficient 3 4 2 Arguments for Application of Presto Version V1 0 Only Presto is really prepared for the modeling of dynamic effects i e for the identification of time delays between input and output variables NeuroSystems in principal assumes a static characteristic surface i e delay free effects from the input variables to the output variables For larger softsensor applications with numerous input variables Presto offers advantages with respect to modeling features and model per formance that can be achieved As a tool from experts for experts Presto offers a lot of functions and user definable parameters and is promising very good results if applied by professionals If the application of INCA c f section 4 is planned anyway the applica tion of Presto
13. suggests itself because both tools work neatly together in a common runtime environment on a separate PC July 9 2009 29 48 Copyright Siemens AG 2009 All rights reserved SIEMENS Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 ModPreCon APC AddOn Products 37361131 4 Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 ModPreCon Although there are a lot of different multivariable control algorithms in the ory e g state space controllers H controllers the model predictive con trollers MPC dominate the field in industry Like suggested by the term model based a dynamic model of process behaviour including all interac tions is used inside the control algorithm to predict future process move ments in a defined time span The control problem is interpreted and solved as an optimization problem The optimal trajectory of the manipulated vari ables MVs minimizes both the sum of future control errors and the sum of future MV moves The following section is a typical example for the comparison of a DCS em bedded lean MPC and an external full blown MPC with online optimiza tion as discussed in general form in section 1 4 of the whitepaper cited above 4 1 Comparison in a Table Model Predictive Control Table 4 1 Product information INCA MPC Ipcos Novel SIMATIC PCS 7 MPC bzw Controller Architecture ModPreCon Software provider IPCOS NV Leuven Belgium Siemens AG
14. variables Visualization of relevance of Selection via genetic algo each input rithms or beam search Modeling of dynamic sys Dynamic model types or con Only feasible using work tems sideration of time delayed arounds series connection of inputs deadtime blocks in front of individual input variables estimation of deadtimes us ing external tools mani pulation of learning data to make them deadtime free System identification Selection of different model Three types of artificial neural types networks Linear transfer functions e Mulitlayer perceptron General non linear e RBF network radial Models GNOMOs basis functions Fuzzy logic e Neuro fuzzy system Partial least squares estimators Prior knowledge about the can be applied in the de Is not necessary but there plant sign is also no way to apply it inside the tool besides the selection of input variables Verification of process mod Comparison of model Comparison of model els output and measured output and measured data in trend curve data in trend curve Individual time slots can Validation data can be be declared to be learn read from a separate ing or validation data data file or selected sto The statistical evaluation chastically from the of models can be based learning data on learning and or valida Animated 3D graphics tion data characteristic surface Further graphical evalua tions scatterplot residual analysi
15. versus PCS 7 Mod PreCon Version V1 0 July 9 2009 6 48 Copyright Siemens AG 2009 All rights reserved SI EM ENS Introduction APC AddOn Products 3 361131 Figure 1 1 Detail from the interactive catalogue of the IA amp DT Mall A amp D Mall Interactive catalog Ent Process control systems He Lj SIMATIC PES 7 V7 0 sor Migration to SIMATIC PCS 7 FE Add ons for SIMATIC PCS 7 ei Information and Management Systems Boc Advanced Process Control res E INCA Model predictive multi variable controller rn Presto Soft sensor for nan measurable quality variables De E Rarlb Expert tool forthe optimization of PID controllers neste E ADCO Adaptive controller dE MATLAB SIMULINK DDE client Online coupling for APC ent I FuzzyCantral Configuration tool for fuzzy logic Eee LI New Neuro Systems Configuration tool for neural networks arm Industry specific applications Version V1 0 July 9 2009 7148 Copyright Siemens AG 2009 All rights reserved SI E M E N S Optimization of PID controllers RaPID by Ipcos versus SIMATIC PCS 7 PID Tuner APC AddOn Products 37361131 2 Optimization of PID controllers RaPID by Ipcos versus SIMATIC PCS 7 PID Tuner Many PID controllers in industry are tuned by trial and error methods or by heuristic rules and the differential action is frequently not considered at all For certain standard control loops like the flow control of fluids with a pro portional val
16. 1 e Orange setpoint e blue CV in the past e red prediction of free CV response if MVs are frozen e green planned optimal CV trajectory e CVs controlled variables e MVs manipulated variables e DVs disturbance variables Figure 4 5 GUI of INCA_Modeler ioli g Project View Window Help alaj xl Dee ees es elal 4 m GIE Lal Overview TectPragct Data Pioperiies Data D epe Data E Den m Diameter PV Average Diameter L3 Inds wall hickness Pu zu c E Sawe Import d Export b Luk copy Pestle Delete Rename Madalz Views Execute Properbes IB Data 4 B Cases 3 ld Models 3 a Abos ave 01 31 E In a modeling project there are folders with data and models A case is a combination of data and model structure the execution of a case is the identification of model parameters from these data Version V1 0 July 9 2009 37 48 Copyright Siemens AG 2009 All rights reserved SI E M E NS Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 ModPreCon APC AddOn Products 37361131 Figure 4 6 Comparison of two models in INCA_Modeler a INCA Modeler FirModel GnsTestModel Project View Window Help Dealer Elsa aaa gt BI 40 20 X DrawingSpeed SP Minutes Version V1 0 July 9 2009 38 48 Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 ModPreCon SIEMENS APC AddOn Products 37361131 Figure 4 7 Parameters
17. 2 4 Functionality INCA PID Tuner alias Ra PID Robust Advanced PID Control Independent of DCS Predefined templates for common PID algorithms by Siemens ABB Honeywell Emerson etc The appropri ate structure has to be manually selected Controller types OPC interface to PCS 7 Operator Station or offline evaluation of measurement data files Data acquisition Test signals e Setpoint step e Manipulated variable step e Ramps e Pseudo random binary noise signals PRBNS Data pre processing Select time slots None Filter data System identification Selection of different model types with without deadtime system order to be selected arbitrarily Prior knowledge about the can be applied in the de plant sign Optimization of PID controllers RaPID by Ipcos versus SIMATIC PCS 7 PID Tuner 37361131 SIMATIC PCS 7 PID Tuner PID function blocks from PCS 7 Standard Library and Advanced Process Library are supported automatically With V7 0 or higher there is also an interface for different but similar function blocks from 3 party libraries Trend curve recorder inte grated in tuner assistant e Setpoint step e Manipulated variable step PTn models only system order is determined auto matically deadtimes are approximated by higher sys tem order IS not necessary but there is also no way to apply it inside the tool Verification of process model Mode
18. 6 2 DPA r3 Chart Edit Insert CPU Debug View Options Window Help Eh m e Gu gi SS EHE l AG BEM M y New Chart Mew Text CFC Library current CFC library a FUZZY NEURO Library FuzzyControl BausteinBausteine Maure MEURO_64K all blocks SIMATIC 57 Meur com runc OB_FUNCT 4 MOVE REGELUNG NeuroSystems BausteiniBausteine Cox All blocks COM_FUNC OB_FUNCT MOVE ig Inputa ourpurs El po aur L ams EEE Bo FROWN MAHE oh QurPUIS I F FE IMPUTZ2 DUTFUTZ F F ED Copyright Siemens AG 2009 All rights reserved D elle d sr pol A H H pi LA NERO 64K FB103 SIMATIC 57 x lt Ill il Blocks Charts M Libraries Find initial letter Press F1 for help Sheet 1 0635 CFC CFETI Neuro Version V1 0 July 9 2009 26 48 Copyright Siemens AG 2009 All rights reserved SI E M E NS Soft Sensors Based on Artificial Neural Networks Presto by Ipcos versus SIMATIC NeuroSystems APC AddOn Products 3 361131 Figure 3 6 Faceplate of function block Neuro 64K SIMATIC S7 MeuroSystem 64K Daten Version 5 1 PRAIRIES HEC M Was ouis a 55 Output oae Oupu3 ES je oes Outputs 061 Outputs i nous a TT 000 uu TE mas 7 04 out a ut aa T om 2 S fa ae Fa in a Input 17 0 00 Input 37 7 0 00 en en E Ao o eme put soon ees Input 20 7 0 00 nn Version V1 0 Ju
19. C AddOn Products 37361131 Table 4 4 Functionality INCA MPC Ipcos Novel SIMATIC PCS 7 MPC bzw Controller Architecture ModPreCon Number of controlled vari unlimited typically 3 20 lt 4 ables CVs can be varying at runtime constant at runtime become smaller Number of manipulated vari unlimited typically 5 20 lt 4 ables MVs frequently not equal to num usually equal to number of ber of CVs CVs can be varying at runtime can be varying at runtime become dibus become smaller ables DVs can be acivaied at intime can be activated at runtime Control zones around set Control zones around set points soft constraints points soft constraints MV limits hard constraints MV limits hard constraints Control targets CVxZone CVxldeal CVxDy SPx SPxDeadBand namic MVxMovePenalty MVxMovePenalty MVxldeal each of them with weight each of them with rank and weight Optimization Online iterative solution of Analytical solution of optimi optimization problem in each zation problem ignoring con sample step considering straints This solution can be constraints and hierarchy of calculated offline based on control targets ranks performance index and proc Algorithm quadratic ess model and delivers a programming QP Solver mathematical formula that requires only a few matrix multiplications for the online calculation of MVs Test signals Generation of special Typic
20. C PCS 7 Add on Catalogue The modularity flexibility scalability and openness of SIMATIC PCS 7 of fers ideal conditions for integrating additional components and solutions into the process control system and completing and extending their func tionality in this way Since SIMATIC PCS 7 was launched on the market we at Siemens as weli as our external partners have developed a host of supplementary compo nents that we refer to in short as PCS 7 add ons The catalogue is available in the internet via the IA amp DT mall https mall automation siemens com DE guest index asp aktprim 0 amp nodel D 10008888 amp lang en amp foldersopen 1303 1300 1 8523 8524 8525 8745 4545 amp jumpto 8745 The responsibility for a PCS 7 add on product generally rests with the ap propriate product manager External SIMATIC PCS 7 partners organize the sale and delivery of their products independently Their own terms and conditions of business and delivery apply In the add on catalogue section Advanced Process Control you will find the software packages shown in Figure 1 1 In the following areas of APC methods the customer in principle has the choice between an add on product and an APC function already included in PCS 7 Optimization of PID controllers RaPID by Ipcos versus PCS 7 PID Tuner Softsensors based on artificial neural networks Presto by Ipcos ver sus SIMATIC NeuroSystems Model based predictive control INCA by Ipcos
21. active for predictive control 4 4 2 Arguments for the Application of INCA For larger MPC applications with more than 4 interacting MVs and CVs the combination of several ModPreCon function blocks with coupling by a disturbance compensation at the joint is principally feasible but INCA will provide better control performance in such cases As a tool from experts for experts INCA offers a lot of features func tions and tuning parameters and is promising very high performance if applied by professionals In the following cases the application of INCA is strongly recommended because ModPreCon does not dispose of the required features Version V1 0 Larger control problems where several MVs have to be driven to the constraints in order to achieve optimal performance Only an online op timization is capable of finding the ideal working point at the intersection of several constraints at runtime Larger control problems where the number of degrees of freedom is varying frequently at runtime because CVs or MVs are switched on off or are hanging at limits Only an online optimization considering con straints can make sure that the mathematically optimal solution of the constrained problem is really found July 9 2009 44 48 Copyright Siemens AG 2009 All rights reserved SIEMENS Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 ModPreCon APC AddOn Products 37361131 Control problems with much more
22. ally a series of step PRBNS test signals based on experiments rough process model Test signals must be gener Test signals can be activated ated by user in manual mode using an additional applica of ModPreCon tion called INCA_ Test Data acquisition using INCA Test using trend curve recorder of CFC Data preprocessing offline Selection of several time e Selection of time slot slots e Low pass filtering Low pass filtering e De trending De trending System identification Numerous model forms to Universal fixed model type be selected by user ARX model of 4 order plus e Finite Impulse Response deadtime for each transfer FIR models channel Version V1 0 July 9 2009 32 48 Copyright Siemens AG 2009 All rights reserved SIEMENS APC AddOn Products Prior knowledge about the plant Verification of process mod els Controller design Handling of nonlinear proc esses Data preprocessing Alarming Version V1 0 Model Based Predictive Control INCA by Ipcos versus INCA MPC Ipcos Novel Controller Architecture State space models semi automatic order selection using Hankel singular values ARX models identified by output error minimization Laplace transfer func tions in continuous time can be applied in the de sign Comparison of model output and measured data in trend curve hot explicitly required Numerous controller parame ters can be adjusted online F
23. ap File Edit View Design Properties ES Noise sensitivity Pro ject 10 X km h 2 fi O kmh u Robustness AI 65 65 amp Acquisition u Parameters T ou K 8 63 X km h Ti 2 06 sec EN Td 0 0094 sec Identification Mr 1 10 20 sn an so 60 70 an an s E time sec E pw Con 4cton Reversed Control Design D c E 3 Legend zn a M Controlled Yariable Se E nr SetPoint Im Disturbance 5 m MF Manipulated Variable IV Load 0 10 zi 30 zu 50 ELI 70 au g time sec 12 12 2002 22 10 2 Version V1 0 July 9 2009 12 48 Copyright Siemens AG 2009 All rights reserved SI E M E N S Optimization of PID controllers RaPID by Ipcos versus SIMATIC PCS 7 PID Tuner APC AddOn Products 37361131 Figure 2 2 Process model in RaPID tr NC 52 26 awastwz Ke Parameters Numerator Denominator k Gain 0 347 egrees z T Delay 3 min Offsets CV Offset 485 D egrees 0 653 MY Offset 44 7 F 0 076 cycles min Integrator t Damping Ratio T Time Constant f Natural Frequency t 2xII f Close Figure 2 3 Selection of different controller types in RaPID Controller settings X Controller template Structure Configuratio Er Emerson Frovox General amp h D aac TDUSUUU Universal SetPoint amp h ee ESCK Dicital Controller Siemens PCS CTRL PID
24. case of mandatory liability e g un der the German Product Liability Act Produkthaftungsgesetz in case of intent gross negligence or injury of life body or health guarantee for the quality of a product fraudulent concealment of a deficiency or breach of a condition which goes to the root of the contract wesentliche Ver tragspflichten The damages for a breach of a substantial contractual obli gation are however limited to the foreseeable damage typical for the type of contract except in the event of intent or gross negligence or injury to life body or health The above provisions do not imply a change in the burden of proof to the detriment of the orderer Copyright 2009 Siemens Industry Sector IA These application ex amples or extracts from them must not be transferred or copied with out the approval of Siemens For questions about this document please use the following e mail address mailto online support automation siemens com Version V1 0 July 9 2009 2 48 Copyright Siemens AG 2009 All rights reserved SIEM ENS Preface APC AddOn Products 37361131 Preface Objective of the Application Besides the APC functions in the SIMATIC PCS 7 APC Library respectively Advanced Process Library there are some more APC software packages in the PCS 7 add on catalogue available on www automation siemens com e INCA Model based predictive multivariable controller e Presto Softsensors for quantities not
25. dd Siemens PCS CONT C amp h Siemens T eleperm Error amp h Yokogawa E TI Integra EE derivative O Yarable double derreative proportional derivative double derreative Controlled arable Controller x cma Version V1 0 July 9 2009 13 48 Copyright Siemens AG 2009 All rights reserved SI E M E N S Optimization of PID controllers RaPID by Ipcos versus SIMATIC PCS 7 PID Tuner APC AddOn Products 37361131 Figure 2 4 Comparison of different control designs in RaPID Compare controllers oj x T 2 um er T cC n im 5 oy LL e ea B ya 3 50 100 150 200 time sec 5 5 4 3 2 1 0 1 Load 50 100 150 200 time sec Po geument BenchMark 2 19 V m 0 958 50 x Step Rise Time Step Settling Time Step Oyvershoat Hide Version V1 0 July 9 2009 14 48 Copyright Siemens AG 2009 All rights reserved SIEMENS APC AddOn Products Optimization of PID controllers RaPID by Ipcos versus SIMATIC PCS 7 PID Tuner 2 3 Illustrations for PID Tuner Figure 2 5 PCS 7 PID Tuner in CFC 37361131 z az E Chart Edit Insert CPU Debug View Options Window Help e x Dia S EE Bc e dS e m rS Ent zl amp amp eEITII 3 Anlage ConPerMon ConPerMonSim TIC201 PID Tuner iol xl Setpoint 100 4
26. directly measurable e RaPID Expert tool for optimization of PID controllers e ADCO permanent adaptive controller e Matlab Simulink DDE Client Online interface for APC e FuzzyControl Engineering tool for fuzzy logic e NeuroSystems Engineering tool for artificial neural networks Some functionalities like e g fuzzy logic or adaptive control can be realized only with an Add on product On the other hand in the areas of PID optimi zation neural networks and predictive control the customer has the choice between an add on product and an APC function already included in PCS 7 The following contribution is intended to support taking the appropriate de cision considering the setting of a task the desired function range set of features and the non functional requirements The illustrations added provide a visual impression of the graphical user in terfaces of the software tools As opposed to the other application notes they are not intended to be a step by step manual for the application of the software More detailed information concerning features and usage of the software tools can be found in the original documentation of the respective products Main Contents of this Application Validity The following main points are discussed in this application note Optimization of PID controllers RaPID by Ipcos versus PCS 7 PID Tuner Softsensors based on artificial neural networks Presto by Ipcos ver sus SIMATIC Neu
27. e 2 sec 1 2 5 10 elc DB no 1 30 1 Design Controller Bode Diagram SUL 1000 1500 2000 mm n T n im 5 E En m 100 oy n n m tai m CL Version V1 0 July 9 2009 43 48 Copyright Siemens AG 2009 All rights reserved SIEMENS Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 ModPreCon APC AddOn Products 37361131 4 4 Hints for Selection of Appropriate Product 4 4 1 Arguments for the Application of ModPreCon Higher availability of runtime algorithm in automation system up to ex ploitation of redundant SIMATIC hardware More easy integration in PCS 7 No software license costs Less engineering costs Look amp feel of ModPreCon are similar to conventional PID controllers Therefore you need less time to get familiar with it and in most cases there is no need to call for external consultants as experts for special ized MPC software packages In general INCA is a tool from experts for experts similar to the other add on products by Ipcos and requires the appropriate time to get fa miliar with the software and the theory behind while the no charge Siemens tool has advantages with respect to usability Summing up the starting prize for a turn key ModPreCon solution is re duced by an order of magnitude compared to an INCA solution This also means that small and medium sized applications that do not allow amortizing a full blown MPC become attr
28. han in equivalent to conventional the central controller of a function blocks inside DCS DCS and moreover can make use Therefore supervision via of redundant AS hardware watchdog is required Version V1 0 July 9 2009 18 48 Soft Sensors Based on Artificial Neural Networks Presto by Ipcos versus SIMATIC NeuroSystems SIEMENS APC AddOn Products 37361131 Table 3 3 Usability INCA Sensor alias Presto SIMATIC NeuroSystems Properties Estimator Windows start menu Via context menu in CFC of Neuro function block or via Windows start menu User guidance in engineering tool Interactive Windows program with numerous menus and numerous user specified parameters Interactive Windows program with simple menus and a small number of user speci fied parameters Operator monitoring and Compact GUI of Presto Standardized PCS 7 face control during operating Online including functions for plate phase input of lab sample results Parameters for the user data block of Neuro function block are supplied by NeuroSys tems tool Transfer of configuration data PrestoOffline creates csv to runtime algorithm configuration file for Pres toOnline Table 3 4 Functionality INCA Sensor alias Presto SIMATIC NeuroSystems Properties Estimator Number of inputs Unlimited lt 100 typically lt 8 Number of outputs Unlimited lt 10 typically lt 4 Copyright Siemens AG 2009 All rights reserved Data acquis
29. ion V1 0 July 9 2009 35 48 Copyright Siemens AG 2009 All rights reserved SI E M E NS Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 ModPreCon APC AddOn Products 37361131 Figure 4 3 Features and faceplates of the three APC interface function blocks s p Controllers v valve control s p Controllers CV xB Timeout 58 10000 Mode Program x TE mw AliveSignal 0 Betp Ape Predictive Mode 3 21 kmolim Cv ENT Communication supervision New operating mode Program Data filtering Central switchover of PIDs cascade with setpoint from APC Outlier detection and alarm Predictive Mode Prediction Bumpless switchover Operation of MPC similar to PID Control Mode enabling switchover Operation of MV parameters controller Program Mode all PIDs switchover J Status bytes for MPC Status bytes alive Polymerization Reactor AC Supervisor nn EX AC Interface Block Pressure Control TF Pressure TF Viscosity AC MA en oc IDEAL 0 05 96 AC CTRL Ideal PV Figure 4 4 MPC graphical user interface INCA View with online visualization of predictions i INCAview Overview Fust 04 23 2002 09 44 96 Last 04 24 2002 1729 36 383 Logide If Tray Version V1 0 July 9 2009 36 48 Copyright Siemens AG 2009 All rights reserved SI E M E N S Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 ModPreCon APC AddOn Products 3736113
30. ition Data preprocessing offline Version V1 0 Offline evaluation of measurement data files Excel Text Access etc Data including time stamps also suitable for dynamic models Import of several data files is supported Comprehensive statistic of raw data Selection of time slots Data filtering De Trending Outlier elimination Resampling of datasets with different sampling rates July 9 2009 Offline evaluation of measurement data files Ascii Text tab delimited in fixed format Data without time stamps because only static models are identi fied Only one file for learning data and optionally a second file for validation data Statistical distribution of learning date incl Mean value and variance Option for normalization of input and output data based on learning data file For further data preprocess ing external tools like MS Excel or Matlab must applied 19 48 Copyright Siemens AG 2009 All rights reserved SI EMEN S Soft Sensors Based on Artificial Neural Networks Presto by Ipcos versus SIMATIC NeuroSystems APC AddOn Products 37361131 INCA Sensor alias Presto SIMATIC NeuroSystems Properties Estimator e Arithmetic operations e Automatic Normalization Selection of relevant input Comparison of models with Combination visualization of variables different structures i e with correlation of input and out different combinations of put signals input
31. l fit Available since V7 0 Controller design Mathematical parameter optimization using simulated scenarios for setpoint follow ing or disturbance rejection allows well defined specifica tion of requirements Simulation of control loop Can be fully parameterized Exact quantitative evaluation of results comparison of different control designs additional frequency domain analysis Manual input at Operator Station and CFC Transfer of controller pa rameters Version V1 0 July 9 2009 P PI PID according to stan dard formula of modulus optimum optimal distur bance rejection Optional detuning of setpoint response Fixed pre defined scenario Simulation available since V7 0 Loading into AS and offline data management of CFC via mouse click 10 48 Copyright Siemens AG 2009 All rights reserved SI E M E N S Optimization of PID controllers RaPID by Ipcos versus SIMATIC PCS 7 PID Tuner APC AddOn Products 37361131 Literature e http www ipcos com cms uploads INCA 20PID 20 Tuner pdf e lpcos User Manual RaPID Jan 2007 e Siemens PCS 7 PID Tuner Online Help V7 0 1 Nov 2007 Version V1 0 July 9 2009 11 48 Copyright Siemens AG 2009 All rights reserved SI E M E NS Optimization of PID controllers RaPID by Ipcos versus SIMATIC PCS 7 PID Tuner APC AddOn Products 37361131 2 2 Illustrations for RaPID Figure 2 1 RaPID user interface U RaPID Skilifti71202 r
32. ly 9 2009 27 48 Copyright Siemens AG 2009 All rights reserved SI E M E N S Soft Sensors Based on Artificial Neural Networks Presto by Ipcos versus SIMATIC NeuroSystems APC AddOn Products 37361131 Figure 3 7 GUI of engineering tool NeuroSystems ey Konfiguration NeuroSystems C Programme SIEMENS NeuroFuzzy Demo Examples NeuroSystems Schicht snl Datei Bearbeiten Zielsystem Lemen Test Ansicht Fenster osla a gt eles vj C Programme SIEMENS NeuroFuzzy Demo Examples MIE PRES r la PA a FRE C Programme SIEMENS NeuroFuzzy Demo Examples NeuroS ystems Schicht snl 2 3D Grafikdarstellung Eingange p 31 Pa es C Programme SIEMENS NeuroFuzzy Demo E xamplesXN euroSystems Schicht snl 3 Kurvenschreiber 11 27 24 7 26 27 28 11 27 30 11 27 32 11 27 34 11 27 36 11 27 38 11 27 40 Bl ttern im Archiv A Dr cken Sie F1 um Hilfe zu erhalten di Start a Explorer NeuroFuzzy Demo Posteingang Microsoft 0 Microsoft PowerPoint Ad E Konfiguration NeuroS Version V1 0 July 9 2009 28 48 Copyright Siemens AG 2009 All rights reserved SI EMEN S Soft Sensors Based on Artificial Neural Networks Presto by Ipcos versus SIMATIC NeuroSystems APC AddOn Products 37361131 3 4 Hints for Selection of Appropriate Product 3 4 1 Arguments for Application of NeuroSystems Higher availability of runtime algorithm in automation system up to ex ploitation of
33. mens AG 2009 All rights reserved SI E M E N S Soft Sensors Based on Artificial Neural Networks Presto by Ipcos versus SIMATIC NeuroSystems APC AddOn Products 37361131 3 1 Comparison in a Table Soft sensors based on artificial neural networks Table 3 1 Product Information INCA Sensor alias Presto SIMATIC NeuroSystems Properties Estimator Software provider IPCOS NV Leuven Belgium Siemens AG IS Erlangen and Boxtel Netherlands http www ipcos be Delivery form External product in add on Siemens product in add on catalogue catalogue Table 3 2 System architecture INCA Sensor alias Presto SIMATIC NeuroSystems Properties Estimator Integration in PCS 7 Separate software tool on Optional software tool in PCS external PC 7 ES Runtime algorithm PrestoOnline as OPC DA CFC ready SIMATIC function Client on Windows PC with block NEURO_ 64K connection to Operator Sta The runtime software does tion requires typically an not require a lot of computing Ipocs DataServer and Sched power but an additional user uler as runtime environment date block for parameteriza A PCS 7 function block with tion corresponding data structure in the OS e g OpAnL is used as an interface in the WinCC OPC Server The runtime software does not require a lot of computing power and can be installed directly on an OS client Availability of software on Windows of runtime software is PC is generally lower t
34. n inia Table 18 3 2 llis HAUG NS TOL PIE conspecta idt tonnes Me san 22 3 3 Illustrations for NeuroSystems nano nnnno nenn nennen 25 3 4 Hints for Selection of Appropriate Product 29 3 4 1 X Arguments for Application of NeuroSystems 2 2002200220002200 nenn nennen 29 3 4 2 Arguments for Application of Presto wwmwmmmanwamwanwamznwwawnza 29 4 Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 MOGOPIOCGOD ES nuu a DE M MAI ME 30 4 1 GColparisohqMd Table 30 4 2 IMIS tEAODIS TOP TIN GA are td dete fee t dnd alas pae itum dint aun UG ed dE 34 4 3 Illustrations for MOoGOdPEeCG ON 40 4 4 Hints for Selection of Appropriate Product 44 4 4 1 Arguments for the Application of ModPreCon eeeeeeeeees 44 4 4 2 Arguments for the Application of INCA nsennensnnnennnnnnnennsnrenrsrrenrsrrereerrererene 44 5 SUMMARY ee nee u Ra Da Un NM EOM genauen 47 6 HISTOIV eerte eisen 48 Version V1 0 July 9 2009 5 48 Copyright Siemens AG 2009 All rights reserved SI EM ENS Introduction APC AddOn Products 3 361131 1 Note Introduction A general overview of the PCS 7 embedded APC functions Advanced Process Control is provided by the White Paper How to Improve the Performance of your Plant Using the Appropriate Tools of SIMATIC PCS 7 APC Portfolio http pcs khe siemens com efiles PCS 7 support marktstudien WP PCS 7 APC EN pdf SIMATI
35. ns AG 2009 All rights reserved SI E M E NS Soft Sensors Based on Artificial Neural Networks Presto by Ipcos versus SIMATIC NeuroSystems APC AddOn Products 3 361131 3 2 Illustrations for Presto Figure 3 1 System architecture for external soft sensor Presto Cups 5 minat on Tine 7 21 2003 AT 07 PM 1 Bas FRE Lab Time POSE SOY PET RESET BAS eg eee RE Learning data PrestoOnline incl GUI gt PC runtime calculations operator monitoring and b Les control OS Clients minis PrestoOffli EY E 5 restoOffline Engineering Fos mm network training Station ES 4 OS LAN Ethernet OS Server redundant Industrial Ethernet Fast Ethernet Version V1 0 July 9 2009 22 48 Copyright Siemens AG 2009 All rights reserved SI E M E N S Soft Sensors Based on Artificial Neural Networks Presto by Ipcos versus SIMATIC NeuroSystems APC AddOn Products 37361131 Figure 3 2 Software structure of Presto The so called DataServer is an Ipcos internal OPC server that is con nected via an OPC delegator to the process interface i e the OPC server of an OS Client Version V1 0 July 9 2009 23 48 Copyright Siemens AG 2009 All rights reserved SI E M E N S Soft Sensors Based on Artificial Neural Networks Presto by Ipcos versus SIMATIC NeuroSystems APC AddOn Products 37361131 Figure 3 3 GUI of PrestoOnline Presto VIA SYNC Output 3 8804E 00 Time
36. o the different phase of a batch process A batch to batch observer and controller provide adaption of setpoints and constraints based on measurements of batch end quality By using existing physical models for heat and energy balances of the reactor the effort required for experimental modeling is reduced consid erably A nonlinear model of the reaction kinetics is deduced from his torical batch data July 9 2009 45 48 Copyright Siemens AG 2009 All rights reserved SI E M E NS Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 ModPreCon APC AddOn Products 37361131 Figure 4 12 Functionality has its prize Functionality Version V1 0 July 9 2009 46 48 Copyright Siemens AG 2009 All rights reserved SIEMENS em APC AddOn Products 37361131 5 Summary Some similarities are obvious in all three comparisons Advantages of PCS 7 embedded APC products Availability is in general higher on the SIMATIC CPU compared to a Windows PC moreover the advantages of redundant hardware can be exploited Costs for the PCS 7 embedded APC products there are no or only small software license fees Usability for the PCS 7 embedded APC products there is less expert know how required look amp feel are similar to conventional automation functions the user is guided Engineering effort the PCS 7 embedded APC products are developed to allow for fast and easy engineering and commissioning the number of
37. of the Ad vanced Process Library will provide dedicated interfaces for external APC software tools The controller runtime soft ware requires a lot of com puting power and must be installed on a separate PC Availability of software on Windows Is equivalent to conven PC is generally lower than in tional controller function the central controller of a blocks inside DCS and DCS moreover can make use of Therefore a conventional redundant AS hardware backup control strategy in side the DCS and supervi sion via watchdog is re quired Table 4 3 Usability INCA MPC Ipcos Novel SIMATIC PCS 7 MPC bzw Controller Architecture ModPreCon Windows start menu Via context menu in CFC of ModPreCon function block or via Windows start menu User guidance in Interactive Windows program MPC configurator with prede engineering tool INCA Modeler with numer fined sequence of three ous menus and numerous working steps user specified parameters Number of parameters to be specified by user is mini mized Operator monitoring and GUI INCA View with numer Standardized PCS 7 Face control during operating ous possibilities for parame plate phase terization Look amp feel similar to PID con Online visualization of predic troller tions Version V1 0 July 9 2009 31 48 Copyright Siemens AG 2009 All rights reserved SIEMENS Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 ModPreCon AP
38. omation and Drives Online Help of PCS 7 APC Library V7 0 SP 1 Nov 2007 e Siemens AG Sektor Industry Online Help of PCS 7 Advanced Process Library V7 1 Mar 2009 4 2 Illustrations for INCA Figure 4 1 System architecture for external full blown predictive controller INCA s p Controllers amp C Supervisor T Ea sess ui APC faceplates B Timeout 58 AC alive AliveSignal 0 C Predictive Mode Control Mode Program Mode DE Suppress Message Few LATE Las Dumme men l Inca suite OS Clients MPC engineering and runtime SW OPC client Engineering Station ES OS LAN Ethernet M tat Puce Bai OS Server redundant Industrial Ethernet Fast Ethernet Version V1 0 July 9 2009 34 48 Copyright Siemens AG 2009 All rights reserved SI E M E N S Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 ModPreCon APC AddOn Products 37361131 Figure 4 2 Software structure of INCA Suite INC Aengine ov Database interface Process interface Process Equipment process The so called DataServer is an Ipcos internal OPC server that is con nected via an OPC delegator to the process inferface i e the OPC server of an OS Client The projekt specific operator user interface is real ized using the APC faceplates on the OS The alternative process connec tion via a database interface is not applied in the context of PCS 7 Vers
39. or Simulation there is a separate elaborate tool called INCA Simulator e Gain scheduling e Trajectory control In APC interface blocks or via Smooth function block Inside Ipcos environment i e outside of DCS and addi tionally with APC interface blocks inside of DCS July 9 2009 SIMATIC PCS 7 ModPreCon 37361131 SIMATIC PCS 7 MPC bzw ModPreCon Automatic conversion to Fi nite Step Response FSR model for controller Is not necessary but there is also no way to apply it inside the tool Comparison of model output and measured data in trend curve automatically in MPC con figurator requires the specification of CV weights and MV move penalties only can easily be verified in side the configurator tool by simulation Model scheduling solu tion template since PCS 7 V7 1 Trajectory control desig nated in ModPreCon function block dedicated modeling and activation of trajectories currently still require applicative ef forts Can be realized with stan dard CFC function blocks e g Smooth Can be realized with stan dard CFC function blocks e g MonAnL 33 48 Copyright Siemens AG 2009 All rights reserved SI E M E N S Model Based Predictive Control INCA by Ipcos versus SIMATIC PCS 7 ModPreCon APC AddOn Products 37361131 Literature e http www ipcos com cms uploads INCA 20MPC pdf e pcos user manual INCAEngine V7 1 Jan 2007 e Siemens AG Aut
40. parameters to be specified by the user is minimized Advantages of Ipcos add on products Version V1 0 Functionality the Ipcos products provide nearly all features that are available from mathematical theory in this context Performance the Ipcos products are state of the art with respect to control performance approximation precision etc Flexibility a large number of tuning parameters allow adapting the Ip cos products very precisely to specified requirements Application area the high performance products by Ipcos can be ap plied even for very large or very difficult applications July 9 2009 47 48 Copyright Siemens AG 2009 All rights reserved SIEMENS buse APC AddOn Products 3 361131 6 History Table 6 1 History Version V1 0 July 9 2009 48 48
41. roSystems Model based predictive control INCA by Ipcos versus PCS 7 Mod PreCon valid for PCS 7 V7 0 SP1 and V7 1 Version V1 0 July 9 2009 3 48 Copyright Siemens AG 2009 All rights reserved SIEM ENS Preface APC AddOn Products 37361131 Reference to Industry Automation and Drives Service amp Support This article is from the internet application portal of the Industry Automation and Drive Technologies Service amp Support Clicking the link below directly displays the download page of this document htip support automation siemens com WW view de 37361131 Version V1 0 July 9 2009 4 48 Copyright Siemens AG 2009 All rights reserved SI E M E N S Table of Contents APC AddOn Products 37361131 Table of Contents Table OF COMENS nenn ana 5 1 IriigorejU eir o piss E 6 2 Optimization of PID controllers RaPID by Ipcos versus SIMATIC PCS 7 PID TUFHOT ee 8 2 1 Comparison Mid Tables ce cares Rester a e meh co PAR Eo Ine ee at 9 2 2 Ilustrauons Tor RaP Diana a ea du otn at 12 2 3 lustrauonsTor PIDSTUPIGESGu ciem Cu ae 15 2 4 Hints for Selection of Appropriate Product 16 2 4 1 Arguments for Application of PID Tuner sn nennen 16 2 4 2 Arguments for the Application of RaPID sensanoenennsennnnnnnsnnsrrerrsrreresrrereerrene 16 3 Soft Sensors Based on Artificial Neural Networks Presto by Ipcos versus SIMATIC NeuroSystems eeeeeee e eeeee rennen nean 17 3 1 Compariso
42. s Data preprocessing online Inside OPC client incl Outlier Can be realized using stan detection peak shaving dard CFC function blocks e g Smooth Alignment with laboratory Bias update module IS missing measurement results Version V1 0 July 9 2009 20 48 Copyright Siemens AG 2009 All rights reserved SI E M E N S Soft Sensors Based on Artificial Neural Networks Presto by Ipcos versus SIMATIC NeuroSystems APC AddOn Products 37361131 INCA Sensor alias Presto SIMATIC NeuroSystems Properties Estimator Information regarding reliabil Confidence intervals for cal are missing ity of calculation results culated output values Alarming Inside Ipcos environment i e Can be realized with stan outside of DCS If needed an dard CFC function blocks additional alarming inside of e g MonAnL DCS can be realized with additional CFC function blocks Literature e http www ipcos com cms uploads INCA 20Sensor pdf e lpcos user manual Presto 2007 e User manual SIMATIC NeuroSystems V5 1 Siemens AG 2008 e Contact product manager NeuroSystems Langer Gerhard Industry Sector IS IN E amp C OC IT PRODUCTS Erlangen e Dittmar R Vergleich von Werkzeugen zur Entwicklung von Soft Sensoren auf der Grundlage k nstlicher neuronaler Netze Studie im Auftrag von A amp D GT 5 B M Pfeiffer FH Westk ste Heide Holstein August 2001 Version V1 0 July 9 2009 21 48 Copyright Sieme
43. te process model that describes the correlation of these variables In literature this approach is called soft sen JI sor virtual online analyzer or property estimator There are several methods to develop soft sensors The best results can be achieved using theoretical process models relying on physical thermody namical and chemical first principles Unfortunately this approach is not feasible in many cases because the cost for theoretical modeling is not jus tified with respect to the expected benefit Empirical modeling based on historical process data requires less effort however it does not always succeed The disadvantage is that such models in soft sensors are valid only in this operation region where process data are available in sufficient amount and quality because extrapolation capa bilities of such model are very limited If the correlation between measurable process values and quality variables to be estimated is strongly nonlinear the application of artificial neural net works for modeling is well established because they don t require to pre specify the exact mathematical structure of the nonlinearity The structure of an artificial neural network roughly resembles the structure of biological brains involving a huge number of neurons and interconnections where the knowledge about the detailed correlation is stored in the connection weights Version V1 0 July 9 2009 17 48 Copyright Sie
44. ve there are empirical values for standard parameter sets For slow controlled processes like temperature control loops an optimization by trial and error takes too much time because the observation of a single step response may need several hours Consequently the application of computer aided controller design tools is winning recognition The systematic optimization of the subordinate PID controllers has to be performed before any supervisory MPC can be ap plied because the slave closed loops are part of the time invariant proc ess model used in the MPC master controller and cannot be re tuned later on The principle sequence of steps for computer aided controller design stays the same from PID to MPC The process is excited with a step of the ma nipulated variable or a setpoint step if there is at least a stable but subop timal controller setting A dynamic process model is estimated from the stored measurement data by the tuning tool i e the process parameters are calculated such that the learning data are fitted optimally in a least squares sense by the model The calculation of the optimal controller pa rameters is based on the identified process model Version V1 0 July 9 2009 8 48 Copyright Siemens AG 2009 All rights reserved SIEMENS Optimization of PID controllers RaPID by Ipcos versus SIMATIC PCS 7 PID Tuner APC AddOn Products 37361131 2 1 Comparison in a Table Optimization of PID Controllers Table 2 1
Download Pdf Manuals
Related Search
37361131_APC_AddOns_en