Graphical user interface system for steam turbine operating conditions
Contents
1. 73 25 01 potentially troublesome operation conditions 5 386 689 2 1995 Bozich et al 60 39 33 5 439 160 8 1995 Marcantonio 228 102 8 Claims 14 Drawing Sheets TACOUISITION O SIMATIC 55 ETHERNET COMMUNICATION LeveL lt 0 OMNI SERVER OECnet PHASE V 8 1 LEVEL TELEGRAM DISTRIBUTOR TIME 4 RaTs REQUIREMENT TELEGRAM E TELEGRAM RECEIVER FACTION LEVEL BLADE TEMPERATURE ESTIMATOR T EVALUATE 8 PROCESS CONTROL MANAGER icc DATA LEVEL PRESENTATION LEVEL FREE i GRAPHICS SHARED MEMORY WINDAGE GUI U S Patent Nov 17 1998 Sheet 1 of 14 5 838 588 AR Sp IA BUM VR Nets ui ere 2 92 7 y i 1 SIMATIC 55 COMMUNICATION LEVEL ETHERNET OMNI SERVER DECnet PHASE V Y 1 LEVEL TELEGRAM TME TELEGRAM WATCH DISTRIBUTOR CONTROL w RECEIVER RQTs RDTS i REQUIREMENT TELEGRAM RAW DATA TELEGRAM ACTION LEVEL PROCESS CONTROL MANAGER BLADE TEMPERATURE ESTIMATOR EVALUATE _ T2. 2 57227 0092 002 0062 NN N N N M UN ANY N NW N Q N 222 Yj 000E 222222 001 2 ZlY DG AA et 2 1 SS AMY eo i VME tif 7 7 4 AGTWHINS PYN 14345 36 913 U S Patent HOLINOW WWHOVI 8313817879 3 1 9661 TT ch 230 NHL 318 39VONIM SNJ3N3IS U S Patent Nov 17 1998 Sheet 10 of 14 5 838 588 ANE N Y 52 E gt 3 NN i WIN Y NW O D x 22222 ao C CC pP CC ZZ A P gt 22 a ENTROPY 5 N N Z JP pU Z 2 2222222 YELL Z Z 2 22 TAT 2 222 AES oa HP MOLLIER DIAGRAM D OA 222222 222222 2222 Z GZ AN Ne WIN AXE 2222 2222222 LA N NN 2 DIAGRAM D NN FILE PARAMETER TURBINE 5 838 588 Sheet 11 of 14 Nov
3. 17 1998 U S Patent m A Z gt Z Y LP P tj ZL 2 4 22220 Yj 000 nk 00 2 12 Z 7946 Z Z ZZ 00 5 2 005E WUE 7 7 009 222 WG 8 ou WWHOVIG 38 2 3 5 838 588 Sheet 12 of 14 Nov 17 1998 U S Patent 00 9c 0 Oc 0 HOLINOW 9667 50 92 0 330 0y S WIL 10 00 NVHOVIO dH L0 318 39VONIM GB ViE LY 96 L0 0 0c or 09 08 0 00 007 009 008 0007 0001 00 009 0087 0002 00cc 00be 009 008 000 00cE 00 009 008E JEG td 9 0 001 00c 005 00 006 009 007 008 006 000 0077 000 005 007 006 009 00 7 0087 0061 000 001 00 00Ec 00 005 009 00 0 945 INIAN HdVu9 31 H3MOd 1fid1f0 WduyG GOr N 3445 L 0 06 0ep 031 3 29707 121 95901 O LILE 790 021 O ONISVO 3 07 11 3 9 2 59 O 9 08 8 O 0103 18 3 3 80 078 511 3unive3dW3l 30 8 L 2 08 0 O Jet Td ONIOVIG 380438
4. Arana 339 Assistant Examiner Hien Vo Attorney Agent or Firm Adel A Ahmed 56 References Cited 57 ABSTRACT U S PATENT DOCUMENTS A graphical user interface provides a real time information 3 873 817 3 1975 Liang 364 492 display for a supervising engineer in charge of turbine 4227093 10 980 Uram etal 290 40 R Operation so that critical parameter values and undesirable 4 576 007 3 1986 Arakawa et al e 60 660 Combinations of operating conditions are readily observed 4 679 309 7 1987 Strickler ces 60 646 and deviations are made apparent so that corrective action 4 764 025 8 1988 Jensen 374 144 can be initiated rapidly An overview of the operating 4 827 429 5 1989 Silvestri Jr 364 494 situation is made more readily apparent by representing the 4 891 948 1 1990 Kure Jensen et al 60 645 operating expansion and compression processes by lines on 4 970 670 11 1990 Twerdochlib 364 571 03 a Mollier enthalpy entropy chart In combination real time 5 267 435 12 1993 Frenkel et al 60 204 parameter values and parameter trends are also presented 5 305 230 4 1994 Matsumoto et al 364 495 Using the Mollier chart information in conjunction with EM 2195 ae UU trend and real time information the supervising engineer 53332040 7 71994 Matsumoto et al T 395 23 can more quickly identify and correct undesirable and 5 353 628 10 1994 Bellows
5. finding the global optimal solution The solutions are then tested using the data test file The solution with the smallest error will be used in the estimation process during the background processl In addition to the current values and trend diagrams the graphical user interface will also able to show the turbine conditions within the steam behavior Mollier diagram This diagram also called a Mollier chart entropy enthalpy diagram or a total heat entropy diagram serves as a familier environment for any thermodynamics engineer and a better representation of the turbine condition with respect to all known critical operating boundaries Therefore this on line turbine condition visualization will better help a user in taking appropriate control actions Generally the GUI process must be initiated by the user It will access values stored by the background process as required The GUI process follows the following steps see the correponding illustration in FIG 4 The Windage Graphical User Interface Module can be initiated independently or from within DIGEST This will automatically initiate the connection to the Shared Memory unit The shared memory unit is basically a routine which manages the access and transfer of data between the GUI and any process outside it which mainly includes a buffer From the front page FIG 5 a the user can select through the TURBINE menu so as to view any of the following turbine windows HP t
6. is claimed is 1 In a system for monitoring steam turbine blade tem perature utilizing measurement parameter values a graphi cal user interface utilizing a computer for displaying a menu so as to allow selection for viewing of any of the following turbine diagram windows turbine overview HP turbine LP1 turbine LP2 turbine any other turbine included within the system 5 838 588 9 wherein for each turbine view windows selectable through said menu are provided including turbine overview actual on line turbine condition on a Mollier diagram and a trend diagram window said turbine overview window displaying a current value of blade temperature and said Mollier diagram and said actual turbine condition on said Mollier diagram being generated automatically by said computer based on thermodynamic calculations and blade temperature estimation by a hybrid artificial neural network 2 A graphical user interface in accordance with claim 1 wherein said turbine overview view window displays other information deemed important for a user in making a decision concerning the control of said turbine 3 A graphical user interface in accordance with claim 1 wherein said trend diagram view window allows the selec tion of up to ten diagrams 4 A graphical user interface in accordance with claim 2 wherein said Mollier diagram is generated by said computer utilizing a routine which will generate a background Mollier grid a
7. process can be described as follows The first step is data construction which basically com bines the data obtained from simulation using water steam cycle analysis and data obtained from the experiments Such analysis is for example included in thermodynamics mod ules within the DIGEST system The water steam cycle analysis is used inside the themodynamic module in the DIGEST system As has been explained the DIGEST moni toring system is currently available in the market through SIEMENS AG Next the data is re formatted such it matches the input format of the ANN The data is then reorganized by sepa rating the data into two separate data files where one is used for training and validation purposes and one for testing purposes Although there is no certain rule for regrouping the available data data should be reorganized such that all 5 838 588 7 operating regions should be well represented In accordance with the present exemplary embodiment 8046 of the avail able data is utilized for training and validation and the rest for testing The ANN structure is a standard multilayer with 1 hidden layer The number of hidden units may vary from 4 to 10 without significant improvement in performance a longer traning period is needed for larger number of hidden units and it may run the risk of overfitting In reference to FIG 3 starting with an initial set of traning parameters including type of optimization algorithm type of
8. temperature thresholding which allow the user to set a certain threshold for activating the warning label and sending an alarm signal to the operator The trend diagram allows the selection of up to ten parameters to be shown at the same time The maximum number of parameters that can be shown is essentially unlimited however any number larger than ten will cause difficulties in viewing the graph itself It has the same feature as feature 2 in the Mollier diagram The exact value within a graph can be found by clicking on the desired point The exact value will be displayed under the corresponding axis From the trend diagram window the user can further analyze the data by selecting the FREE GRAPHICS which will give the user access to the complete data base This component is provided within the DIGEST system The GUI display process will access the necessary data from the Shared Memory with the exception of the FREE GRAPHICS routines which will access data from the data base through the data server While the invention has been described by way of exem plary embodiments various changes and modifications will suggest themselves to one skilled in the art who becomes familiar with the invention For example the choice of parameters made herein can be changed as a matter of choice or convenience These and like changes are contemplated to be within the scope and spirit of the invention which is defined by the claims following What
9. 8 60 8 0 0c 07 09 08 001 Oct 007 097 087 00c Oce Orc 08e 00E Oct Ove 09 08 00r Voy RACE ocr Jeq U3d Jeq utg shod 45 318 39VONIM WYHOVIO 0 GW 0 0 0c roe roe Ob FOr 08 09 09 09 08 08 08 00 Oct 07 091 08 00 Oce 09 08 00E Oct Ove 09 08 007 Oey 001 097 087 00c Oce Ove 09c 08c Oct Ove 09 08E 00r 067 3NI8Hfil 001 09 08 00 Oce Ove 09 08 00 Oct Ove 0 08 007 Och HdVu9 3LVOdN 80 10 101 3 0 109110 O 04445 9 82 05 0211 3 9 02107 O 3 770 81 O 2 98001 O 3uniYu3dH3l 941983 3 2507 1 3 0 Ed SNH 3 88 607 01 31 30 id 2 2 82 01 45077 Td HALJNVHVd ANGNI 3i SN3N3IS PS 9I3J 5 838 588 1 GRAPHICAL USER INTERFACE SYSTEM FOR STEAM TURBINE OPERATING CONDITIONS In the operation of steam turbines as for turbo generators it is important that operating parameters be kept within defined limits for proper and safe operation includ ing start up and shut down phases Unsafe operation can have grave consequences for personal injury and material damage Reference is hereby made to an application by the present inventors being filed concurrently herewith and entitled A METHOD FOR BLADE TEMPERAT
10. ED MEMORY MANAGEMENT FIG 4D TREND DIAGRAM mee HP LP TREND GRAPH SIMILAR STRUCTURE FOR LP4 LP2 DATA DRAWING ROUTINE m 4 GRAPH DATA MET ee ka SE e pluie dus ES U S Patent Nov 17 1998 Sheet 6 of 14 5 838 588 s eJ a a e eo 4 lt c co lt r 4 ca uL cc LI cc 1 c L a uJ ul U S Patent Nov 17 1998 Sheet 7 of 14 5 838 588 FIG 5B Tf3 405 04 C Pms 401 87bar Tms 403 23 C Ti 401 76 C CEO J e Pex 406 49bar Teh 420 31 C IP CONDENSER FIG 5C Tblade 412 96 C Pin 419 9Bbar Tcw 416 57C Tin 410 69 C Teb 410 74 C TY Tm Tci 403 56 C Peh 152 71 Teh 420 09bar U S Patent Nov 17 1998 Sheet 8 of 14 5 838 588 FIG 5D Tblade 414 76 C Pin 410 30bar Tin 403 12 C Tcu 407 55 C N 403 12bar Tcb 408 67 C Tciz415 95 C Tco 409 20 C Peh 160 58rpm Teh 403 06bar 5 838 588 Sheet 9 of 14 Nov 17 1998 WVHOVIO TION dH 5 AdOWLNS 78 8 08 BL 9LvLet 0 88 9 wo Oce oo a co gt m cN cu V N WE y cO M Y N Y lt d NY N N Y W Y Y 1 1 WS N A 0062
11. P and LP turbines in accordance with the present invention will pro vide the operator with an estimation of the blade temperature at the respective turbine stages The interactive user inter face herein disclosed displays the real time value a trend graph of these values and the respective states within the Mollier diagram Supervisory recommendation may be deduced from the estimation and other available measure ment values In the following examples of the windage phenomenon are given In the HP turbine as there is no steam flow through the turbine following a trip the extent of energy transfer depends on the pressure and the steam density in the turbine At a full load trip the corresponding high cold reheater pressure will initially be present In order to avoid impermissible heating by windage losses an adequate pres 5 838 588 3 sure decay or a certain cooling steam flow is required The expansion lines in the Mollier diagram indicate the advan tage of a sufficient HP turbine flow after full load rejection to zero load The operator is much better informed by such a figure On line visualization of the expansion compression lines is especially beneficial for other parts of the turbine which are subject to overheating due in the present particular case to the windage phenomenon For heating steam turbines when the control valves for example in the cross over line for the two lower heaters are closed the LP turbine requi
12. RAINING __ TRAINING __ MATHEMATICAL i i 1 I I 1 RESULT 773 SCHEME PU wx PARAMETERS pb L 21 U S Patent Nov 17 1998 Sheet 2 of 14 5 838 588 LAYER S OF HIDDEN UNITS FIG 2B P3 MATHEMATICAL CALCULATION 4 ANN MODULE ee MATHEMATICAL 1808 CALCULATION 2 LAYER S OF HIDDEN UNITS U S Patent Nov 17 1998 Sheet 3 of 14 5 838 588 FIG 3 WATER STEAM CYCLE ANALYSTS PROGRAM GENERATE DATA EXPERIMENTAL DATA COMBINE DATA REFORMAT AND ORGA NIZE TRAINING AND TESTING DATAFILE ANN MODULE TRAINING AND VALIDATION MODIFY TRAINING PARAMETERS OR THE SATISFACTORILY ANN STRUCTURE ANN MODULE TESTING SATISFACTORY U S Patent Nov 17 1998 Sheet 4 of 14 5 838 588 FIG 4A MAIN MENU FRONT PAGE MOLLIER TREND TURBINE IMAGE SHARED MEMORY DATA SERVER DATA BASE FIG 4B FRONT PAGE HP LP1 Da PARAMETER LABELS DRAWING ROUTINE A UL Pe e E a i E GE GSE CE E E E EEE ERE M eirca SHARED MEMORY MANAGEMENT EEE A ae ey ee U S Patent PARAMETER LABELS PARAMETER LABELS Nov 17 1998 Sheet 5 of 14 5 838 588 FIG 4C MOLLIER DIAGRAM a HP LP MOLLIER GRAPH SIMILAR STRUCTURE FOR LP1 LP2 BACKGROUND DATA DRAWING ROUTINE ROUTINE NEU n DATA SHAR
13. TOS 8872858 O 15 313 0 3113 SN3N3IS HG 91J 5 838 588 Sheet 13 of 14 Nov 17 1998 U S Patent 0y 5 0086 0 HOLINON 9667 0755 0 ET 330 105 5 311 S 10 Ov 58 10 00 NVH9VIO Id EG 40 0 06 007 05 00 0Se 00 0 008 097 006 OGG 009 069 00 08 08 05 00 067 00c 05 00 OSE 007 09 006 086 009 0 amp 9 007 08 008 06 007 06 00 06 00 OSE 007 097 006 066 009 099 00 08 008 9 191 9 831 cdl Id dH 318 JOVONIM 06 007 OST 00c 06C 00 OSE 007 067 006 05 009 0 amp 9 007 06 008 2 031 Jed uod 0 Jed Utd 3 9 0 0 00 Oct 0 097 087 00 0 6 ovv 097 087 006 05 06 HdVH9 311090 3804 109110 9996 707 03395 3 86 317 031 3 2 00121 3 68 517821 3 96 017 221 3p vu3di3l 9NISYO 0 951 3 50 del O BEDE 7 00 508 1 30V18 2 62 901 O 00 097 04 WVHOVIG ANIGUNL 3113 SN3N3IS IS 9I3 5 838 588 Sheet 14 of 14 Nov 17 1998 U S Patent 00 00 80 ZB 966 St 00 Oc bG 0 HOLINOW 80 330 IHd Ov 5 3811 8 10 00 3 10 N381 dH 85 0 Oe G 900 QT 89 98y 10 96
14. URE ESTI MATION IN A STEAM TURBINE Ser No 08 764 381 whereof the disclosure is incorporated herein to the extent it does not conflict with the present application Typically in steam turbo generator operations the tur bine was operated around full power or when the demand for power was insufficient it was shut down Particularly in operation as part of a large power grid operation at less than full load may be required Under such conditions complex patterns of temperature pressure steam wetness reheating expansion and compression may occur possibly resulting in excessive turbine blade temperature Such conditions may spell blade failure with possibly disastrous consequences Thus monitoring operation under conditions where the intake steam pressure is at or lower than the output pressure are of practical importance Background material is avail able in books such as W W Bathie Fundamentals of gas turbines John Wiley and Sons 1996 and H Herlock Axial flow turbines Fluid mechanics and thermodynamics Butterworth London 1960 Good mathematical models for simulating the steam behavior in a turbine in its entire operating domain are not readily available especially concerning periods in which the main steam pressure is near or lower than the exhaust pressure During such periods the fluid flow behavior is very complex because the radial component of velocity become significant as compared with the axial veloci
15. US005838588A United States Patent ro 1 Patent Number 5 838 588 Santoso et al 4 Date of Patent Nov 17 1998 54 GRAPHICAL USER INTERFACE SYSTEM 5 479 358 12 1995 Shimoda et al 364 492 FOR STEAM TURBINE OPERATING 5 640 176 6 1997 Mundt et al 345 146 CONDITIONS FOREIGN PATENT DOCUMENTS 75 Inventors Nugroho Iwan Santoso Cranbury 0 389 132 of 0000 European Pat Off N J Walter Zorner Baiersdorf OTHER PUBLICATIONS Germany Patent Abstracts of Japan vol 004 No 072 M 013 27 73 Assignee Siemens Corporate Research Inc May 1980 amp JP 55 035112 Toshiba Princeton N J Comparing Display Integration Strategies for Control of a Simple Steam Plant Edlund et al 2 Oct 1994 Proceed 21 Appl No 768 047 ings of the Int l Conference on Systems Man and Cyber netics San Antonio pp 2686 2691 22 Filed Dec 13 1996 Omega Engineering Inc The Data Acquisition Systems 51 itas 05 15 00 vol 29 p B 1 to b 18 1995 USS SSS 304 528 22 E Turbine Back Pressure Identification And Optimization 364 468 01 364 528 25 With Learning Neural Networks Mathur et al Advances in 58 Field of Search 364 557 468 03 Instrumentation And Control vol 45 No Part 01 1 Jan 364 468 04 468 01 528 22 528 25 345 145 1990 pp 229 236 146 340 428 915 919 920 921 970 Primary Examiner 1 ouis M
16. activation function number of hidden units error thresholds the training process is started The optimization algorithm used is a standard technique available in various optimization or Neural Network textbooks See for example Hertz A Krogh and Palmer Introduction to the theory of neural computation A lecture notes volume in the Santa Fe Institute Studies in The Sciences of Complexity Addison Wesley Publishing Company July 1991 and D Rumelhart J L McClelland and the PDP Reseach Group Parallel distributed processing Explora tion in the mocrostructure of cognition Volume 1 Foundations MIT Press Cambridge 1987 Several techniques were investigated in conjunction with the present exemplary embodiment including gradient descent and few conjugate gradient techniques Faster con vergence is obtained by applying the one variation of conjugate gradient techniques If the system satisfactorily converges such that the vali dation error thresholds are satisfied then the ANN param eters connection weights and unit s threshold values are stored for testing If the system does not converge then the training parameters must be modified until a solution is obtained The processes above may be done repeatedly since it is generally known that the system may converge to different solution with different initial condition and training param eters Obtaining siginificant number of solution may increase the possibility
17. al time parameter values and parameter trends are also presented Using the Mollier chart informa tion in conjunction with trend and real time information the supervising engineer can more quickly identify and correct undesirable and potentially troublesome operation condi tions In accordance with an aspect of the present invention a system utilizes a hybrid ANN artificial neural network algorithmic based scheme for estimating the blade tempera ture from other measurements which are commonly avail able The commonly available measurement values are herein utilized The training data for the ANN includes both data generated by mathematical model and by experiment The invention will be better understood from the follow ing detailed description in conjunction with the drawing in which FIG 1 shows a windage module architecture in accor dance with the invention FIG 2 shows an artificial neural network based scheme for blade temperature estimation in accordance with the invention FIG 3 shows a training procedure for an artificial neural network in accordance with the invention FIG 4a to 4d shows graphical user interface structures applicable in conjunction with the invention and FIG 5a to 5j shows graphical interface views applicable in conjunction with the invention During the operation of the steam turbine heating due to windage must be maintained within allowable limits by the operating mode The windage modules for H
18. d on the turbine operating conditions Several scenarios can be pre determined for each specific turbine For example no load full load and low load during slow shutdown start up and load rejection The monitoring cycle should be adjusted automatically for different conditions depending on their criticalities and the respective display may be arranged to pop up to assist the operator The windage module basically has two main processes the background process and the interactive display process The background process is responsible for obtaining the necessary parameter values calculating the blade tempera ture at a predefine rate and recording the relevant informa tion into the appropriate shared memory and data base The interactive display process will show the necessary or 5 838 588 5 requested information graphically at any point of time The process rate is limited by the minimum amount of time required before all measurements stabilize and will vary based on the severity of the turbine condition Operation near the critical blade temperature may require faster pro cess rate Before the monitoring process the ANN must be trained The training sub structure is responsible for producing the appropriate weights and parameters that will be used in the monitoring module This process is done off line and is not controllable through the GUI interface The network is trained using the simulated data obtained by computing the
19. e a satisfactory result In one method the ANN is used directly to derive operating blade temp values 10 15 20 25 30 35 40 45 50 55 60 65 2 In a accordance with another aspect of the invention a hybrid approach 5 measured values are utilized A subset of for example 4 parameter values is used for training the ANN and another subset of for example 3 values is used for performing a calculation for another intermediate parameter Using the intermediate parameter and one of the 5 measured values a blade temperature is calculated In accordance with still another aspect of the invention the user interface provides a real time information display for a supervising engineer in charge of turbine operation so that critical parameter values and undesirable combinations of operating conditions are readily observed and deviations are made apparent so that corrective action can be initiated rapidly While graph plots of parameters can be readily presented such a format generally does not readily provide an overall picture of the state of the turbine with regard to the distribution and combination of temperature pressure steam wetness or superheat and turbulence effects In accordance with the present invention an overview of the operating situation is made more readily apparent by representing the operating expansion and compression pro cesses by lines on a Mollier enthalpy entropy chart In combination re
20. e operation so that critical parameter values and unde sirable combinations of operating conditions are readily observed and deviations are made apparent so that corrective action can be initiated rapidly said interface providing an overview of an operating situation made more readily apparent by representing operating expansion and compres sion processes in real time by lines automatically generated on an automatically generated Mollier enthalpy entropy chart based on thermodynamic calculations and blade tem perature estimation by a hybrid artificial neural network together with real time parameter values and parameter trends
21. estimated temperature using the analytical means for the expected normal operating domain and actual data obtain from field experiments The experiments concentrate on generating data in specific low steam flow conditions such as shutdowns loss of loads and start ups This arrangement is expected to be able to estimate the blade temperature for the entire turbine operating ranges Minimal inputs to the estimator are the real time measurement values of the pres sure of the main steam temperature of the main steam pressure of the third stage and exhaust pressure Additional inputs can be optionally provided and evaluated The background process will obtain measurement data calculate the blade temperature and other necessary values and store those values in appropriate locations The process sequences are as follows Request the necessary measurement data to Acquisition Level through the Communication using DEC S5 protocol and Administrative Levels telegram distributor Receive measurement data from data acquisition system Simatic 5 Siemens PLC The request is propagated through the ethernet network communicated using the S5 DEC protocol and managed by the tele capture within the admistrative level The list of the measure ment parameters include Pms Pressure of main steam bar Tms Temperature of main steam C P1 Steam pressure before blading bar Ti Steam temperature before blading C P3 Pressure at the th
22. he database 24 it can be accessed easily by all levels 6 Presentation level The presentation level provides a graphical user interface which allow the users to view all the necessary information in several different fashions that is current values trend diagram and Mollier diagram It consist of the Windage Graphical User Inteface 26 Free Graphics 28 and shared memory 30 for storing the intermediate parameter values needed for the user interface The free graphics is an indepen dent graphical tool for plotting any parameter values stored in the data base This tool is developed as a part of the original DIGEST system The information is presented in several layers starting with the main windage screen which will mainly show the blade temperatures The subsequent layers will show the detail conditions for each turbine section These layers will provide information on all parameter values which are relevant to the operator for making appropriate decisions concerning the turbine operation Further detail on the process within this level 15 provided in the following sec tions detail in the next section A development screen is optionally provided for accessing some internal module and system parameters or processes however principally because of security reasons this feature may preferrably be omitted in an actual working version The monitoring process may not always be necessary to cycle at the same rate at all times it should depen
23. ined ANN based on the current input values c Using the two intermediate values the current blade temperature is then calculated using the equation Equa tion 2 below T3 Eqn 2 Tolade j In this manner a separation is maintained between the mathematically unknown model from the known model In this manner the complexity and nonlinearity within the black box ANN model is reduced Moveover this also helps in reducing the ANN model dependence on specific turbine parameters This improves the accuracy and robust ness of the overall estimation scheme including generali zation between different turbines This allows the method to retain flexibility such as in the alteration of intermediate parameters in the light of new knowledge which also applies to input parameters Such adaptability is herein contemplated The blade temperature estimation and other measurement parameters are then stored in two different places the Data Base and intermediate Shared Memory a All values are stored in the Data Base through the Data Server Ps Pex b Values needed for display within the GUI are also stored in a temporary Shared Memory These values are then available for reading by the GUI process FIG 3 shows the general traning process which appli cable to the ANN module either in the direct approach or the hybrid approach The only difference is in the input parameters as indicated in the background process The
24. ird stage Pex Exhaust pressure after reheater bar Peh Exhaust pressure before reheater bar Teh Exhaust temperature before reheater C Tcb Bottom casing temperature C Tcu Upper casing temperature C Tei Inside casing temperature C Tco Outside casing temperature C N Rotational speed RPM Pout Output power MW Preprocess incoming data into the desired format interpreter This process basically reads the incoming data string and reformat it to a standard ASCII format Store data in the intermediate files for futher processing The estimator will calculate the blade temperature value using the measurement values The input measurment values used for estimating the blade temperature at least for the HP turbine are 1 Pressure of the main steam Pms 2 Temperture of the main steam Tms 3 Pressure at the third stage P3rd and 4 Exhaust pressure Pex 5 Rotational speed 5 15 20 25 30 35 40 45 50 55 60 65 6 One approach directly estimates the blade temperature using a straightforward 3 layer ANN FIG 2 a The second approach uses a hybrid technique FIG 2 b by decompo sition of the intermediate parameters where a One intermediate parameter T3 is calculated analyti cally using Eqn 1 Pms P3 where is a given constant related to a specific turbine size b Another intermediate constant n will be calculated by the tra
25. lers PLC 2 of the type Siemens Simatic 5 Documentation on Simatic 5 is available from Siemens Industrial Automation Its capabilities include signal sampling A D conversion limited computation executing sequence process action cycle timing and open communication functions It is used in this context as a data acquisition device where it samples the measurement data at a predetermined rate digitizes it and transfers the data through the ethernet network asynchronously 2 Communication level This level basically is the com munication server 6 which manages the transfer of information between the network and the DEC Digital Equipment Corporation digital workstation machine s The standard DEC module that handles the com munication issue is called Omni Server DECnet PhaseV The processes within the DEC which manage the the data transfer are indicated by DEC SS 8 and S5 DEC 10 DEC S5 manages the data transfer from the adminstrative level to the S5 and S5 DEC manages data transfer from the 55 to the adminstrative level 3 Adimistration level An administration level of control handles the data request from the windage process control by propagating the request in the right format to 8 communication level which is done by a telegram distributor module 12 It also manages the incoming data in a certain format and forwards the data back to the process control for storage This is done by a telegram receiver module 14 Other functio
26. nd then automatically overlay real time data derived from measurement parameter values and blade temperature utilizing said estimation by a hybrid artificial neural net work 5 A graphical user interface in accordance with claim 2 wherein said trend diagram view window allows the display of an exact value at a desired point within a graph by clicking on said desired point whereby said exact value will be displayed under the corresponding axis 6 A graphical user interface in accordance with claim 2 wherein said trend diagram view window allows a user to 10 15 20 25 30 10 further analyze data by selecting FREE GRAPHICS which will give access to the complete data base 7 A graphical user interface in accordance with claim 4 wherein said Mollier diagram view window has the capa bility for any and all of zooming within the enthalpy entropy graph just by cre ating a box with the mouse enclosing the desired region displaying an instant mini trend diagram which can be activated by clicking at the corresponding parameter value table box by way of a Mollier option interface providing user options to personalize viewing parameters and providing temperature thresholding which allows a user to set a certain threshold for activating a warning label and sending an alarm signal to an operator 8 A graphical user interface for providing a real time information display for a supervising engineer in charge of turbin
27. ns include 10 15 20 25 30 35 40 45 50 55 60 65 4 managing the buffer capacity de log 16 self check ing process watch dog 18 and several timers clocks for interrupt purposes time control 20 Self checking process is mainly to check the status of all processes within the system and re boot the system if necessary 4 Action level The action level controls the continuous background process and computation These include the initiation of data request sending RQTs manage ment of incoming data RDTS data storage all com putation processes and storage of results A more detail description of this level can be found in the next section This level may also include the output man agement which test the validity of the computation result In this scheme the results of the hybrid artificial neural network ANN estimator are always compared to the result of the analytical module This verification is required to detect possible bad results which are usually caused by input values which are far away from all samples that had been presented during the ANN training period Large discrepancies may indicate that further re training is in order 5 Data level The data level handles all processes con cerning data storage and access It includes the data server 22 and data base 24 access to the data base must be done through the data server 22 Once the data is stored in the right format into t
28. res cooling steam to hold within permissible limits the tempera ture rise caused by windage in the last stage In this operating mode the steam in the LP turbine absorbs energy resulting from the windage losses which predominate sig nificantly within the last stages In general the windage module will follow the system architecture used in a system known as the DIGEST system DIGEST is a modular monitoring system for power system plant developed by the KWU FTP activity of Siemens Aktiengesellschaft Simens AG a corporation of Germany DIGEST features a modular system architecture which can be divided into six different levels which will be explain briefly below The module components are written in C with much flexibility in building any structure of choice The proposed windage module system architecture is shown in FIG 1 The first two levels are already available as part of DIGEST Modifications were done to the admin istrative and data levels Modifications in both the commu nication and data levels include parameter specification which is needed for requesting the module specific data through the data bus and for creating the data server and data base The main windage module development is done mainly at the action and presentation levels As indicated in FIG 1 the six levels in the windage module are 1 Acquisition level This level manages the data acqui sition process which comoprises several program mable logic control
29. ty component The available simplified mathematical models for simulating the steam behavior during normal loading typically do not perform properly when the intake pressure is near or lower than the output pressure In new large steam turbines temperature measuring devices are installed at the respective stages of the HP and LP casings These measurements provides an indication to the operator or supervising engineer in charge whenever the blade temperature exceeds its limit The need for blade temperature monitoring for smaller and older turbine as well as a more practical and cost effective ways than installing temperature probes has led to a need herein recognized for a practical system for estimating in real time and monitoring turbine blade temperature during operation The present invention is intended to be practiced prefer rably with the application of a programmable computer In accordance with an aspect of the invention a method for blade temperature estimation in a steam turbine utilizes measurement values including pressure and temperature at locations other than directly at the blades principally at the input and output stages Initially blade temperature is simu lated by using a water steam cycle analysis program as well as by directed experiments An artificial neural network ANN is trained by presenting the measurement values and the blade temp values In a present exemplary embodiment it is found that 4 values provid
30. urbine LP1 turbine LP2 turbine 10 15 20 25 30 35 40 45 50 55 60 65 8 or any other turbines of applicable For each turbine there are three view windows that can be selected through the DIAGRAM menu turbine overview FIGS 5 b 5 d mollier diagram FIGS 5 e 5 g or trend diagram window FIGS 5 1 5 j The turbine overview window gives the current value of the blade temperature as well as other information which may be important for the user to make any decision con cerning the control of the turbine The Mollier diagram is generated based on the standard thermodynamic calculation available on any thermodynamic text book such as the afore mentioned books A routine is herein used which will generate the background Mollier grid and then overlay the expansion data which are calcu lated from the current measurement values on top of the grid For example such a routine is available from Siemens AG in VISUM a user manual Version 3 October 1992 Several features which built into the Mollier diagram window include 1 Capability to zoom within the enthalpy entropy graph just by creating a box with the mouse enclosing the desired region 2 Instant mini trend diagram which can be activated by clicking at the correponding parameter value table box 3 Mollier option interface provide ways to personalize the viewing parameters to the user preferences It also provide
Download Pdf Manuals
Related Search