Automated Analysis Of Digital Relay Data Based On Expert .
Automated Analysis of Digital Relay DataBased on Expert SystemX. Luo, Student Member, IEEE, and M. Kezunovic, Fellow, IEEEAbstract—Modern digital protective relays generate variousfiles and reports which contain abundant data regarding faultdisturbances and protection system operation. This paperpresents an expert system based application for automatedanalysis of digital relay data. In this application, forwardchaining reasoning is used to predict expected protectionoperation while backward chaining reasoning is employed tovalidate and diagnosis of actual protection operation. AnEMTP/C based digital relay model with capability of insertionof user-defined errors and generation of files and reports isdeveloped. The analysis capability of this application is testedusing the relay model.Index Terms—relay operation, relay files, data analysis, expertsystem, relay modelI. INTRODUCTIONWITH the development of computer and communicationtechnologies, more and more intelligent electronicdevices (IEDs) such as digital protective relays (DPRs), digitalfault recorders (DFRs), sequence of event recorders (SERs)and remote terminal units (RTUs) of supervisory control anddata acquisition systems (SCADAs) are used in power systemsubstations. They supply abundant data related to monitoring,control and protection of power systems. By analysis of thesedata, very useful information can be generated for systemoperators, protection engineers and maintenance staff [1].Sophisticated software tools are required to perform theanalysis so that the benefits of these data can be maximized.Since the 80’s, expert systems were proposed as a promisingtool for analysis of power system data. Various expert systembased applications such as DFR data analysis, alarmprocessing, and circuit breaker monitor data analysis havebeen reported in the literature [2], [3], [4]. However, most ofthese applications can not perform detailed analysis ofprotection system operation because they generally utilize datafrom DFRs, SERs and RTUs, which only contain limitedinformation about protection system behavior.In recent years, digital protective relays have gained thecapability to generate files and reports which contain detaileddata about power system fault disturbances and correspondingresponses of protection system components. These datainclude samples of analog currents and voltages, statuses ofThe work reported in this paper is funded by NSF I/UCRC called PowerSystems Engineering Research Center (PSERC) under the Project T-17 titled“Enhanced Reliability of Power System Operation Using AdvancedAlgorithms and IEDs for On-Line Monitoring”.X. Luo and M. Kezunovic are with the Department of ElectricalEngineering, Texas A&M University, College Station, TX 77843-3128, USA(e-mails: xuluo@ee.tamu.edu, kezunov@ee.tamu.edu).protection elements and control elements of relays, statuses ofcontacts of relays, communication channels and circuitbreakers. Automated analysis of these data enables protectionengineers to quickly and accurately evaluate operationperformance, identify design deficiency and incorrect settingsand trace component malfunctions [5].This paper presents research results related to developmentof an expert system based application for automated analysisof digital relay data. Section II introduces data contained in thefiles and reports generated by digital relays. Section IIIpresents the conceptual strategy of the analysis. Section IVdescribes an implementation of the application. Section Vintroduces an EMTP/C based digital relay model andpresents a case study to demonstrate the features of theapplication by analyzing files and reports generated by therelay model. Section VI draws the conclusions of this paper.II. DIGITAL RELAY DATAModern digital relays are capable of generating variousfiles and reports, each of which may contain a specificcategory of data. Generally, oscillography data contain therecords of what a relay “sees” during disturbance events.Setting data specifies how the relay is configured. Fault datapresents disturbance information and phasor parameterscalculated by the relay for its decision making. Sequentialevent data reveals how the relay and associated protectioncomponents actually respond to the disturbance events. Thesefour categories of data are introduced as follows.A. Oscillography DataOscillography data generated by the fault recording functionof a digital relay are usually contained in oscillography files inCOMTRADE format [6]. Secondary voltages and currentscoming into the relay are recorded as analog channels whilestatuses of both external contacts and internal states of therelay can be recorded as digital channels by users’ selection.B. Setting DataSetting data contained in a setting file specify configurationparameters of a relay. Usually setting data configures the relayat three levels: selecting protection and control elements,deciding how the selected elements are logically combined,and setting operating parameters of each selected element.C. Fault Disturbance DataFault data contained in a fault report include fault type, faultlocation and voltage and current phasors during pre-fault andfault periods. They are calculated by a relay and used for itsdecision making.
D. Sequential Event DataSequential Event Data contained in an event report are timestamped logic operands in chronological order. It containsmost of the information through which the external behavior ofa relay and its associated protection system components andthe internal states of the relay can be observed. According toour investigation, for some types of relays, not all logicoperands that are important for analysis are reflected in theirevent reports. This problem can be solved if users select theseoperands to be recorded in the oscillography files.Besides the above four categories of data, performancespecification data such as average operating time for Zone 1 ofa distance relay and average operating time of a circuit breakerare also important to our analysis. They are usually containedin user’s manuals or may be obtained from other IEDs.sources: the local relay, remote relays and other fault analysisapplications based on advanced algorithms and techniquessuch as expert systems, Neural Networks and SynchronizedSampling. References [2], [7], [8] provide details of theseadvanced algorithms and techniques for fault analysis.To decide the information source to be used in the analysis,we have the following assumptions.1) The disturbance information obtained from fault analysisapplications based on advanced algorithms and techniques ismore accurate than that produced by relays.2) The disturbance information produced by the relay whichindicates the fault is in its Zone 1 is more accurate than thatproduced by the relay which indicates the same fault is in itsoperation zones other than Zone 1.Based on above assumptions, the logic for choice of sourceof disturbance information is illustrated in Fig. 2. Currentlythis part of logic is not included in our application because weIII. CONCEPTUAL STRATEGY OF THE ANALYSISassume the disturbance information is obtained from an expertThe analysis of relay data is based on comparison of system based DFR data analysis application [2], [9].expected and actual protection operation in terms of statusesWith disturbance information, relay settings andand corresponding timings of logic operands. If the expected performance specification available, the expected statuses andand actual status and timing of an operand are consistent, the timings of active logic operands are inferred by forwardcorrectness of the status and timing of that operand is validated. chaining rules. The results are regarded as hypothesis ofIf not, certain failure or missoperation is identified and protection operation. The actual statuses and timings ofdiagnosis will be initiated to trace the reasons by the use of operands which are obtained from the oscillography file andlogic and cause-effect chain.event report are the facts of p is initiated, it is loadedinto CLIPS inference engine through the windows framework.C. Expert System Reasoning ProcessThe reasoning for prediction of expected protectionoperation is a forward chaining process [11]. Fig. 3 illustratesthe reasoning process, which only details the operation ofground distance elements. The time delay parameters such asdTSUPN, dTPKP P Z and dTOP P Z, which are used to infer thetiming relations, are obtained from relay settings andperformance specification of the relay and associated circuitbreakers.Fig. 3. Reasoning process for prediction of expected protection operation
The reasoning for validation and diagnosis of statuses oflogic operands is performed in two stages. Fig. 4 illustrates thereasoning process. In the first stage, the validation ofcorrectness of statuses of logic operands and diagnosis of thedirect reason for incorrect statuses is performed at all the stepsof the protection operation chain. The validation is based onthe existence and non-existence of hypothesis and fact of anoperand status. If both the hypothesis and the fact exist, thecorrectness of the operand status is validated and there is nodiagnosis information. If the hypothesis exists but the fact doesnot exist, a symptom will be identified and the direct reasonfor the symptom will be diagnosed. In the second stage, thefinal reasons for symptoms identified in the first stage will betraced in a top-down manner by relating together the directreasons for symptoms found in the first stage, which is abackward chaining reasoning process [11].The reasoning process illustrated in Fig. 4 is based onexamination of the existence of a fact if the correspondinghypothesis exists, which aims to deal with such symptoms: Astatus of a logic operand should have existed but it does notexist. There is also a counterpart of the reasoning process,which aims to deal with such symptoms: A status of a logicoperand should have not existed but it exists.The operating speed of protection elements and associatedcircuit breakers is evaluated by examining the timings ofstatuses of logic operands.With the validation and diagnosis information of statuses oflogic operands and operating speed of protection elementsavailable, whether the relay is tripped by the expected elementis further examined and the diagnosis is performed.V. CASE STUDYIn this section, we use a case study to demonstrate thefeatures of the application.A. EMTP/C Based Relay ModelIn order to generate the relay data to be analyzed by theapplication, a digital relay model is developed using C language and MODELS language of ATP program [12]. Therelay model is integrated into a substation model previouslydeveloped using ATP program. The C code realizes therelay function as a foreign model and the MODELS code actsas the interface between the relay function and the ATPprogram which describes the substation model. The relaymodel has the following features.1) Inputs of the relay model include up to three channels ofnode voltages and six channels of branch currents as well asdigital signals such as circuit breaker statuses and blocksignals. Outputs include up to six channels of relay trip signals.Such I/O capability can handle one and a half breaker schemeand single-pole tripping.2) The relay model has four protection elements: PhaseDistance, Ground Distance, Phase Instantaneous Over-Currentand Ground Instantaneous Over-Current. Each element can beindependently enabled or disabled. The distance elements havethree forward protection zones and one reverse protection zone.3) The settings for the relay model are automatically readfrom a setting file at the beginning of the simulation. An eventreport and an oscillography file in COMTRADE format areautomatically generated at the end of the simulation. The usercan select any of the pre-defined logic operands to be recordedin the digital signal section of the oscillography file.4) Users can insert pre-defined errors into the relay modelwhich can cause failures and missoperation. This featurefacilitates testing of the application for relay data analysis.B. Fault ScenarioFig. 5 illustrates the I/O connection of the relay model withthe substation model. The relay DR01 is protecting theoutgoing line L1 which is connected to the substation by oneand a half breaker scheme. The relay takes node voltages onBus B1 as voltage inputs and branch currents through circuitbreakers CB1 and CB2 as current inputs. The two circuitbreakers are controlled by the relay by means of three-phasetripping scheme. The connection between the relay and the twocircuit breakers is simulated by timers within the relay. Thestatuses of breakers are also monitored by the relay.Fault location, fault type and fault inception time can bearbitrarily set on line L1. TABLE I lists the fault informationused in this case study.Fig. 4. Reasoning process for validation and diagnosis of status of logicoperands
Fig. 6 Oscillography file generated by the relay modelFig. 5. I/O connection of the relay model with the substation modelTABLE IFAULT INFORMATIONFault TypeA-BFault Location82 % (Zone 2)Fault Inception Time0.400 secondC. Expected Protection OperationThe relay and associated circuit breakers should respond tothe fault according to relay settings and performancespecifications. TABLE II lists the major characteristics ofexpected protection operation.TABLE IIEXPECTED PROTECTION OPERATIONOperated Element(s)Phase Distance Zone 2Relay Trip Time0.500 secondCircuit Breaker Opening Time0.532 secondCurrent interruption TimeNo latter than 0.548 secondD. Actual Protection OperationIn order to demonstrate the analysis capability of theapplication, we have deliberately introduced some errors in therelay model. TABLE III lists those errors.TABLE IIIUSER-INTRODUCED ERRORS1.Incorrect setting of characteristics of pickup of Phase DistanceZone 2 Element2.Faster opening of Circuit Breaker 1 than performancespecification by one cycle period3.Slower opening of Circuit Breaker 2 than performancespecification by one cycle periodAfter a simulation lasted for 1 second, the relay generates anoscillography file and an event report. Fig. 6 and Fig. 7 showthe oscillography file and event report displayed in the GUI ofthe application respectively. It should be noticed that theoscillography shows the waveform of line currents of line L1,which is the sum of branch currents through circuit breakersCB1 and CB2.E. Analysis ReportThe analysis report is displayed in the dialog shown in Fig.8. It includes relay information, validation information anddiagnosis information. The validation information section listsFig. 7 Event report generated by the relay modellogic operands whose status is as expected and protectionelements whose operating speed is as expected. As shown inthis section, Phase IOC Element operated to make the relaytrip. The circuit breakers opened because of the relay trip.Three abnormities were identified and diagnosed as shown inthe diagnosis information section.Because Phase Distance Zone 2 Element should haveoperated but failed to operate, it was the Phase IOC Elementinstead of Phase Distance Zone 2 Element that made the relaytrip. From such information, we may know that Phase IOCElement functioned correctly as a backup for distance elements.Since the operating time delay of Phase IOC Element was setto be 0.1 second longer than that of Phase Distance Zone 2Element, the relay trip and opening of circuit breakers weredelayed nearly 0.1 second. The reason for failure of operationof Phase Distance Zone 2 Element was the incorrect setting ofits pickup characteristics.Circuit Breaker 1 opened faster than expected by 0.016second while Circuit Breaker 2 opened slower than expectedby 0.016 second.All the three user-introduced errors are identified anddiagnosed, which proves the correctness of the analysis.
VII. REFERENCES[1]M. Kezunovic, C.C. Liu, J. McDonald, L.E. Smith, Automated FaultAnalysis, IEEE Tutorial, IEEE PES, 2000.[2] M. Kezunovic, P. Spasojevic, C. Fromen, D. Sevcik, “An expert systemfor transmission substation event analysis”, IEEE Trans. PowerDelivery, vol. 8, no. 4, pp. 1942-1949, October 1993.[3] S. D. J. McArthur, J. R. McDonald, S. C. Bell, “Expert systems andmodel based reasoning for protection performance analysis”, ArtificialIntelligence Applications in Power Systems, IEE Colloquium on, April20, 1995.[4] M. Kezunovic, Z. Ren, G. Latisko, “Automated monitoring and analysisof circuit breaker operation”, IEEE Trans. Power Delivery. (Accepted,In press).[5] D. Costello, “Understanding and analyzing event report information”,technical paper, Schweitzer Engineering Laboratories, Inc., Pullman,WA, 2000, Available: http://www.selinc.com/techpprs.htm.[6] IEEE Common Format for Transient Data Exchange (COMTRADE) forPower Systems, IEEE Standard, 1999.[7] S. Vasilic, M. Kezunovic, “An improved neural network algorithm forclassifying the transmission line faults”, IEEE PES Winter Meeting,New York, Jan 2002.[8] M. Kezunovic, B. Perunicic, “Automated transmission line faultanalysis using synchronized sampling at two ends”, IEEE Trans. PowerSystems, vol. 11, no. 1, pp. 441-447, February 1996.[9] X. Luo, M. Kezunovic, “Fault analysis based on integration of digitalrelay and DFR data”, IEEE PES General Meeting, San Francisco, CA,June 2005.[10] J. C. Giarratano, CLIPS User’s Guide, Version 6.20, NASA Lyndon B.Johnson Space Center, Houston, TX, 2002.[11] J. Giarratano, G. Riley, Expert Systems Principles and Programming,PWS Publishing Company, Boston, 1994, pp. 158-165.[12] L. Dube, MODELS in ATP, Language Manual, 1996, VIII. BIOGRAPHIESFig. 8 Analysis report generated by the applicationVI. CONCLUSIONSBased on the discussion in this paper, conclusions are drawnas follows:1) Files and reports generated by digital relays containabundant data about the external and internal behavior ofrelays. They are very useful for detailed diagnosis ofprotection system operation.2) Expert systems are powerful tools for protectionengineers to develop intelligent applications for data analysis.3) Forward chaining reasoning and backward chainingreasoning have their own strength. Combination of the twomakes expert system applications more efficient.4) MODELS language of ATP program combined withother high level languages such as C can model verycomplicated logic systems such as multifunctional digitalrelays.Future work on the improvement of the relay data analysisapplication includes two steps. First, the knowledge base of theexpert system will be expanded to enable analysis of largequantity of relay data and interactions of several relays if pilotprotection schemes are involved. Second, one or more faultanalysis applications based on advanced algorithms andtechniques will be integrated to provide accurate disturbanceinformation.Xu Luo (S’05) received his B.E. and M.E. degreesfrom Xi’an Jiaotong University, Xi’an, China, bothin electrical engineering in 1999 and 2002respectively. He has been with Texas A&MUniversity pursuing his Ph.D. degree since August2002. His research interests are power systemprotection, substation automation, and artificialintelligence applications in power systemmonitoring, control and protection.Mladen Kezunovic (S’77, M’80, SM’85, F’99)received his Dipl. Ing. Degree from the University ofSarajevo, the M.S. and Ph.D. degrees from theUniversity of Kansas, all in electrical engineering, in1974, 1977 and 1980, respectively. He has beenwith Texas A&M University since 1987 where he isthe Eugene E. Webb Professor and Director ofElectric Power and Power Electronics Institute. Hismain research interests are digital simulators andsimulation methods for equipment evaluation andtesting as well as application of intelligent methods to control, protection andpower quality monitoring. Dr. Kezunovic is a registered professional engineerin Texas, and a Fellow of the IEEE.
analysis so that the benefits of these data can be maximized. Since the 80’s, expert systems were proposed as a promising tool for analysis of power system data. Various expert system based applications such as DFR data analysis, alarm processing, and circuit breaker monitor data analysis
Nissan Micra - OWM 4734 Nissan Micra - OWM 4734 CPU E6 H33750 1 Battery 2 Ignition switch 3 Fusible link holder 4 Fuse box 5 Fuse and relay box a) accessory relay b) blower motor relay 6 Intelligent power distribution module a) starter relay b) cooling fan hi relay c) cooling fan lo relay d) fuel pump relay e) ignition relay f) ECM relay
38 - Check voltage output to fuel pump @ relay set 39 - Check voltage input to relay set from ignition switch 40 - Check voltage input to fuel pump relay of relay set 41 - Check input voltage to control relay of relay set 42 - Check input voltage to control relay of relay set (continued)
Power MOSFET Relay Specifications (G6D Relay) The following specifications apply to G6D Relays mounted in a G70D Relay Terminal and not the G6D Relay itself. Coil Ratings (per G6D Relay) Note: 1. The must-operate voltage is 75% or less of the rated voltage if the relay is mounted upside down. 2. Rated current and coil resistance were measured .
Results are provided for actual relay testing. Part I: Distance Relay Tests (Texas A&M University) Distance relay testing can evaluate relay performance, calibrate relay settings, and identify system conditions that could cause unexpected relay operation. Developing a relay testing methodology requires consideration of how to model the power .
Relay data and relay setting management Management of relay versions and models World‘s largest library with relay setting data models Library expandable by customer & via relay setting model generator Result import from different test device software: Omicron, Doble, Programma, etc.
Relay AE RLY-538 5 v 178 ohm Resistor DPDT 270 ohms ½ watt Control Tortoise with 5 volt Relay-12 volts DC 12 volts DC Relay On Off To Tortoise Relay “stick” contacts “On” Button powers relay, “stick” contact holds it on “Off” button bypasses relay coil so it drops out Can have any no. of
Relay Computer Harry Porter, Ph.D. Portland State University April 24, 2006. 2. 3 V. 4 Double Throw Relay. 5 V Double Throw Relay. 6 Four Pole, Double Throw Relay. 7 V Four Pole, Double Throw Relay. 8. 9 Schematic Diagrams. 10 Schematic Diagrams Assume other terminal is connected to ground. 11 Schematic Diagrams
Install Your Relay Switch Do not use this sensor in hazardous (classified) locations or life safety applications. Step 1: Mount your Relay Switch. Mount your Relay Switch with screws through the flange holes, secure it with Velcro or simply place it on a flat surface. Step 2: Connect your Relay Switch to AVTECH's Light Tower & Relay Adapter .
