AUTOMATED FAULT ANALYSIS USING ADVANCED INFORMATION TECHNOLOGY FOR DATA INTEGRATION AND INFORMATION EXCHANGE M.

Transcription

1 AUTOMATED FAULT ANALYSIS USING ADVANCED INFORMATION TECHNOLOGY FOR DATA INTEGRATION AND INFORMATION EXCHANGE M. Kezuovic* T. Popovic D.R. Sevcik A. Chitambar Texas A&M Uiversity TLI, Ic. CeterPoit Eergy Etergy Services Summary: This paper addresses some ew approaches to data itegratio iformatio exchage that are made possible due to the use of advaced Iformatio Techology (IT). As a result, some traditioal applicatios such as fault aalysis ca be sigificatly ehaced automated. Paper first describes how the fuctioal ehacemet automatio of fault aalysis may result i faster system restoratio, better quality of power delivery, more comprehesive prevetive maiteace actios. The paper illustrates how i the competitive utility eviromet this leads to improved service ultimately better customer satisfactio while maitaiig eve reducig operatioal cost. Several examples related to automated fault aalysis are described as follows: disturbace aalysis, aalysis of Circuit Breaker (CB) operatio, fault locatio aalysis usig sparse field data. Keywords: Automated Fault Aalysis-Fault Locatio- Modelig simulatio-maiteace-power Quality- Iformatio Techology. INTRODUCTION Deregulatio, liberalizatio, restructurig privatizatio i the utility idustry are simultaeously happeig aroud the globe. They represet differet approaches to makig the utility idustry more cost effective resposive to customer eeds. As a result, two importat treds are recogized: the itroductio of more automated fuctios the use of more advaced iformatio techology. The fial aim is to provide better service to the customers to reduce the operatig cost. This paper illustrates how the existig fault aalysis approach ca sigificatly be ehaced through the use of advaced iformatio techology [1]. I this cotext, the iformatio techology relates to the techology for commuicatios, processig, itelliget systems, computer etworks, databases, user iterfaces, etc. First, the applicatio fuctios ca be improved to iclude broader view of the aalysis icludig data from multiple itelliget electroic devices (IEDs). This leads to a comprehesive data itegratio resultig i efficiet data processig. Oce the relevat iformatio is extracted, it ca be distributed to multiple users resposible for maitaiig the quality reliability of the eergy U.S.A. supply service. All of this ca easily be automated usig iformatio techology services embedded i the applicatios. To illustrate the expected beefits of the use of iformatio techology to support fault aalysis automatio through data itegratio iformatio exchage, several implemetatios carried out by the authors are preseted. The examples iclude: comprehesive fault power quality disturbace aalysis, ehaced just-i-time aalysis of circuit breaker operatio, improved fault locatio aalysis usig modelig, simulatio field data. [2-4]. The paper first gives the backgroud of the ehaced fault aalysis approach based o data itegratio iformatio exchage. The use of iformatio techology to automate the aalysis based o data itegratio iformatio exchage is covered ext. Coclusios refereces are give at the ed. ENHANCED FAULT ANALYSIS BASED ON DATA INTEGRATION AND INFORMATION EXCHANGE Traditioal fault aalysis is based o the use of Digital Fault Recorders (DFRs). The records are brought to the Master Statio located at the Egieer s office, the the aalysis is doe maually goig through the records oe at the time. Depedig o the triggerig methods used to record field data, DFRs may be recordig both faults other similar evets activatig the triggers such as power quality disturbaces, switchig trasiets, spurious oise iterfereces, etc. This makes the aalysis rather tedious sice may of the opeed data records may ot cotai iformatio relevat to fault aalysis. I additio, data from DFRs is rather limited providig aalog cotact sigals from a give substatio with o correlatio to the data recorded by other Itelliget Electroic Devices i this other substatios across the system. The latest developmets i the substatio automatio area have eabled the use of multiple IEDs o the same local Area Network (LAN) while all the devices i a give substatio ca be accessed usig the same commuicatio protocol [6]. I additio, the Iteret services are extesively used to provide system-wide exchage of data iformatio. This allows oe to * Departmet of Electrical Egieerig, Texas A&M Uiversity, College Statio, TX , U.S.A.

2 iclude data from multiple IEDs i defiig implemetig the fault aalysis fuctio. A ew approach featurig data itegratio iformatio exchage is illustrated i Fig. 1. Data may be obtaied from several types of IEDs such as digital fault recorders (DFRs), digital relays (DRs), power quality (PQ) meters, remote termial uits (RTUs), sequece of evet recorders (SERs), programmable logic cotrollers (PLCs), circuit breaker moitorig (CBM) devices, etc. Ehaced fault aalysis is a example of the applicatio that relies o this approach. It may be oted that the aalysis is somewhat broader tha traditioally defied i that it icludes aalysis of both fault power quality disturbaces as well as detailed aalysis of operatio of breakers relays. Applicatios User Iterface (for example, Ehaced Fault Aalysis) Iformatio Exchage Data Itegratio Digital PQ DFRs RTUs SERs PLCs Relays Meters Fig. 1. Data itegratio iformatio exchage CBMs It may be oted that selected IEDs i the figure provide detailed iformatio about: Power quality disturbaces Fault type locatio Operatio of protective relays Operatio of circuit breakers This iformatio has to be extracted from the recorded data provided by separate IED types itegrated for this purpose. The collectio of data, its itegratio processig eed to be doe automatically. Oce the data is processed the obtaied iformatio eeds to be packaged i appropriate reports to serve several differet groups i the utility: Maagers resposible for reliability efficiecy of the system operatio Protectio egieers resposible for correct operatio of the relays related fault clearig equipmet Maiteace crews resposible for performace of the circuit breaker operatio Customer service represetatives resposible for quality of power supply fulfillmet of power delivery cotracts Dispatchers resposible for locatig the faults restorig the system operatio All the metioed groups ca beefit from the data collected by various IEDs give i Fig. 1 as log as there are meas for automatig the data itegratio, processig, aalysis as well as exchagig the aalysis reports, historical data user requests. This ehaced fuctioality is made possible through the use of the IT solutios readily available o the market, appropriately imbedded i the metioed applicatios. The followig sectios give a overview of several advaced applicatios that have used IT services to perform the automatio user iterfacig services. AUTOMATED DISTURBANCE ANALYSIS Fault Aalysis The mai goal of this aalysis is to detect the fault, idetify the fault type, assess the performace of relays, commuicatio schemes circuit breakers ivolved i the fault clearig. A solutio that has bee istalled is used to illustrate how the IT techology has bee embedded with the automated fault aalysis to provide a itegrated solutio [2]. The desig requiremet for this solutio is cetered o the eed to coect DFR Master Statios (MSs) located i multiple service ceters usig the Iteret techology. I additio, each service ceter has a variety of MSs produced by differet vedors, each coected to several DFRs from related vedor. The fial solutio is show i Fig. 2. The aalysis eables reachig the followig coclusios: Fault type locatio Performace of relays, commuicatio chaels circuit breakers Occurrece of ferroresoace coditios autoreclosig sequeces Master Statios Expert System Result Processig Broadcaster Services Waveforms & Reports Reports DFR File Coversio Sigal Aalysis Cetralized Database User Iterface Cliet Server Fig. 2. Automated fault aalysis usig DFRs MSs from multiple vedors. Typical aalysis report display is show i Fig. 3. The followig iformatio is provided i each report: Origial field recorded data Results of the aalysis Substatio cofiguratio data Historical data report search optios

3 The use of IT imbedded with the applicatio allows the followig ehaced user fuctios: Retrievig the data reports usig web browser services Dissemiatig the aalysis results through fax modem, pager, broadcast services Maitaiig the database user iterface settigs through straightforward meas This solutio has show some major beefits whe compared with the traditioal approaches. The compariso is show i Table I. Table I. Compariso of traditioal ehaced fault aalysis approaches Characteristics Traditioal Ehaced Mode of aalysis Maual Automatic Data storage Dedicated, device specific Cetralized database, uiversal Dissemiatio of reports Maual, hardcopy Automatic, , fax, web Istrumets used DFRs /or DRs Ay IEDs User iterface Dedicated, device specific Uiversal, device trasparet Power Quality Aalysis The mai goal of the aalysis is to assess the PQ disturbaces automatically to idetify the oes that are caused by faults (voltage dips) [3]. Iputs: Voltage Waveforms Data Format Coversio Outputs: Detectio Selectio of the Disturbace Type Fourier Wavelettrasform Based Feature Extractio Fuzzy Expert System for Detectio Classificatio Fig. 4. Disturbace detectio classificatio The flowchart of the proposed solutio for PQ evet detectio classificatio is show i Fig. 4. The submodule Data Format Coversio coverts the iputs from a specific recordig device or simulatio package format ito a commo data format comprehesible to other modules of the software. The Fourier Wavelet-trasform Based Feature Extractio module obtais uique features pertiet to specific evets Fuzzy Expert System for Detectio Classificatio module reaches a decisio regardig detectio classificatio, as discussed ext [3]. FFT wavelet-aalysis based feature extractio. The followig eight distict features iheret to differet types of power quality evets have bee idetified: the Fudametal Compoet ( V ), Phase Agle Shift ( α ), Total Harmoic Distortio ( THD ), Number of Peaks of the Wavelet Coefficiets ( N ), Eergy of the Wavelet Coefficiets ( EW ), Oscillatio Number of the Missig Voltage ( OS ), Lower Harmoic Distortio ( TS ), Oscillatio Number of the rms Variatios ( RN )[3]. Fuzzy expert system for detectio classificatio. The core of the rule set of the implemeted fuzzy expert system is illustrated as follows: a) Detectio: For detectio, oe rule is used as follows: Rule 1: if THD is A 2 or PS is B2 or V is C3 or V is C 1 the DETECT=1 b) Classificatio: fiftee rules are used as follows: Rule 1: V + 1 is A 4 N is F1 OS is G1 the IMPULSE=1 Rule 2: V is A1 or V + 1 is A1 the INTERRUPTION =1 Rule 3: V is A6 or V + 1 is A 6 the SWELL=1 Rule 4: V is A5 PS is C1 PS + 1 is C1 EW + 1 is D1 { TS + 1 is H 2 or [ TS + 1 is H4 & TS + 2 is H1 ]} the SWELL=1 Fig. 3. User display i automated aalysis systems

4 Rule 5: V + 1 is A5 { PS is C2 or PS + 1 is C2 } the SWELL=1 Rule 6: V + 1 is A 2 the SAG=1 Rule 7: V + 1 is A3 { PS is C2 or PS + 1 is C2 } the SAG=1 Rule 8: V + 1 is A3 { PS is C1 PS + 1 is C1 } { THD + 1 is B1 or [ THD + 1 is B2 OS + 1 is G 4 ]} the SAG=1 Rule 9: V + 1 is A3 PS is C1 PS + 1 is C1 OS is G 2 THD + 1 is B2 THD + 2 is B2 THD + 3 is B2 the NOTCH=1 Rule 10: V + 1 is A3 N is F2 OS is G 2 the NOTCH=1 Rule 11: V + 1 is A 4 PS is C1 PS + 1 is C1 THD is B 3 THD + 3 is B1 { OS is G 4 or OS + 1 is G 4 } the TRANSIENT=1 Rule 12: V + 1 is A 4 TS + 1 is H3 TS + 2 is H3 TS + 3 is H3 OS + 1 is G 4 the HARMONIC=1 Rule 13: THD + 1 is B4 THD + 2 is B4 THD + 3 is B4 OS + 2 is G 4 the HARMONIC=1 Rule 14: TS + 1 is H 4 TS + 2 is H 4 TS + 3 is H 4 OS + 2 is G 4 the HARMONIC=1 Rule 15: If RN is K1 the FLICKER=1 I the above rules, A B C D F G H K are the membership fuctios for the iput patters, where the commo trapezoidal triagular fuctios are used [3] The fuzzy partitios the correspodig membership fuctios ca be obtaied based o both the statistical studies expert s kowledge. Opiios from operators ca be coveietly icorporated ito the system i practical applicatios. The output for the detectio part is the variable Detect whose value reflects the credibility that certai disturbace exists. The outputs for the classificatio parts are fuzzy variables Flicker, Impulse, Iterruptio, Swell, Sag, Notch, Trasiet, Harmoic whose values represet the degree to which the evet belogs to each of these categories. The type of the evet selected will be the oe with the largest membership. I cases where two or more types of disturbaces have the same largest membership value, all of them will be selected for further aalysis. Evaluatio studies demostrated a correct idetificatio rate of 99%.The correspodig characterizatio algorithms ca be selected for extractig more accurate pertiet parameters. I the cotext of fault aalysis, if the sag is characterized as caused by a fault, this coclusio triggers a exteded fault aalysis that searches for the locatio of the fault, as described later o i the fault locatio aalysis sectio. i Circuit Breaker Operatio Aalysis The mai objective is to aalyze circuit breaker (CB) operatio performace each time it operates. Detailed aalysis of the breaker operatio is eabled by moitorig of additioal cotrol sigals related to CBs with a extesive rule-based expert system [4]. This aalysis also idetifies several breaker operatio features thus poitig to potetial failures i future. The system architecture basic software modules are show i Fig. 5 [4]. The system architecture is typically split betwee two locatios: the substatio where data are recorded durig breaker testig the cetral repository where data from all tested breakers are gathered. Rochester record Coversio Comtrade record SP SP settigs file Extracted sigal parameters O-site aalysis (Cliet) Off-site aalysis (Server) ES settigs file Fig. 5. Fuctioal descriptio of the aalysis software I the particular case, Rochester recorder is used for moitorig the CBs. After the data format coversio, the aalysis software performs sigal processig (SP) i order to extract several features parameters relevat to the operatio of CBs. This module extracts pertiet sigal parameters from recorded sigals usig advaced processig methods such as: wavelet aalysis, Fourier aalysis digital filterig. The extracted parameters are the processed by a rule-based expert system (ES), which is a decisio makig system emulatig reasoig of a huma expert. Aroud oe hudred rules are used i this particular solutio. The cliet part of the applicatio is used o-site (i substatios). The server applicatio is istalled i the cetralized office, i additio to the aalysis, maitais the cetralized database provides a extesive web based user iterface. The server is typically coected to the corporate itraet. There is a optio to upload DFR files to the server from ay computer coected to the compay itraet. Web applicatio (Fig. 7) eables easy maipulatio of data reports as well as additioal tools for maual aalysis by experts. I the cotext of the exteded fault aalysis, the CB aalysis ca be triggered each time the relay or DFR iitiate/detect a operatio of the breaker. I that case, the CB moitorig system ca provide additioal details about the CB operatio ca alert both protectio egieers maiteace crews about related details. ES Evet report Rules file Database Web applicatio

5 Fig. 6. CB aalysis web-based user iterface FAULT LOCATION ANALYSIS USING SPARSE FIELD-RECORDED DATA The proposed approach makes use of the waveform matchig based methodology. The two key cocepts, amely the sparse data waveform matchig are illustrated as follows [5]. Sparse data are the data obtaied from recordig devices sparsely located at various substatio locatios. Examples of recordig devices may iclude digital fault recorders (DFR), digital relays, or other itelliget electroic devices (IED). The data captured by recordig devices may iclude aalog quatities such as voltage curret waveforms digital (cotact) quatities such as breaker status relay operatio status. Both aalog digital quatities may be useful for locatig the fault. To solve the fault locatio problem utilizig sparse data, the waveform matchig based approach may be used. The model of the power system is utilized to carry out simulatio studies. The matchig is made betwee the voltage curret waveforms captured by recordig devices those geerated i simulatio studies. The fault is searched through the system model by utilizig a iterative search process. The search process may cosist of the followig steps. First a iitial fault locatio is assumed. Secod, the simulatio studies are set up accordig to the specified fault. Third, the simulatio studies correspodig to the specified fault are carried out utilizig appropriate simulatio tools. Fourth, simulated waveforms of iterest are obtaied. Fifth, the simulated waveforms are compared with recorded oes, the matchig degree of the simulated recorded waveforms is evaluated by usig appropriate criteria. Sixth, the iitial fault locatio is modified accordig to predefied strategy, the the process proceeds to the secod step cotiues. The above steps are iterated util the simulated waveforms that best match the recorded oes are produced. The fault locatio will be determied as the oe specified i the simulatio model used for geeratig the simulated waveforms that best match the recorded oes. The fault locatio problem ca be formulated as a optimizatio problem: Fid the value of x R that miimize: c f Nv N i k= 1 ks kr f ( x, R ) = { V V } + { I I } (1) or maximize f x, R where f f ( f c f c ( x, R f k= 1 ) = f ( x, R ) (2) ) : defied cost fuctio f x, R ) : defied fitess fuctio. The larger the f ( f value of the fitess fuctio, the better the matchig of the simulated recorded phasors, better the solutio is. x : the fault locatio, ks R f : fault resistace Vks V kr : durig-fault voltage phasors obtaied from short-circuit simulatio studies from recorded waveforms respectively. I ks I kr : durig-fault curret phasors obtaied from short-circuit simulatio studies from recorded waveforms respectively. k : idex of the voltage or curret phasors N v N i : total umber of the voltage curret phasors respectively. It is oted that the largest possible fitess value defied by equatio (2) is equal to zero ca be reached if the phasors obtaied from simulatio studies exactly match those obtaied from recorded waveforms. Therefore, the best fault locatio estimate would be the oe that maximizes equatio (2). It has bee show that the fitess surface obtaied for various fault locatio resistace values is ot regular cotais saddle kr f

6 poits local maximum poits. It is rather difficult to use the gradiet-based method to fid the global maximum poit. Exhaustive search through every possible solutio may also be too time-cosumig to be practical. accordig to the pre-fault breaker status. If both the prefault breaker status the curret of a brach are moitored, the recorded curret overrules the breaker status for determiig the service status of the brach. (2) After the system topology is updated, the PSS/E load flow study is carried out to obtai the pre-fault phasors of the related bus voltages brach currets. The the phasors derived from the recorded waveforms are sychroized by rotatio with referece to the phasors obtaied by the load flow study. (3) Based o the faulty braches give by DFR Assistat the system topology, the software derives the list of total faulty brach cidates for posig faults. (4) The GA algorithm based approach is activated for searchig the fault locatios. CONCLUSIONS Based o the discussio, the followig is cocluded: Data itegratio iformatio exchage eables ehaced fault aalysis May differet utility groups may beefit from the ehaced fault aalysis Iformatio techology is essetial for automatig the aalysis, data report retrieval, user iterfacig Fig. 7. Overall architecture of the software modules The GA based optimizatio approach is good at fidig the globally optimal solutio avoidig the local optima. The ature of the fitess fuctio prompts the attempt of the GA based optimizatio method as described ext [5]. The overall architecture of the developed software is show i Fig. 7. DFR Assistat [7] coverts the DFR raw data ito COMTRADE format [8] obtais the fault type faulty braches based o idividual DFR recordig. Based o such iformatio, the phasors of the moitored voltage curret quatities, pre-fault breaker status ca be obtaied. Based o the topology data a umber of measuremets available, the fault locatio software determies the type of the fault locatio algorithm to be utilized. If special cases like two-ed, three-ed, oe-ed algorithms apply, the fault locatio ca be obtaied directly based o the related voltage curret phasors without ruig ay PSS/E [9] simulatio studies. If the case belogs to the geeral sparse-data case, the the followig steps are to be take. (1) The curret phasors /or the pre-fault breaker status ca be utilized to tue the system topology (update the static system model). Oly the service status of the brach is updated. If the magitude of the pre-fault curret phasor of the moitored brach is smaller tha a pre-specified value, the the brach is cosidered as beig out of service. Brach status is also updated REFERENCES [1] 0.H]XQRYLü&&/LX-0F'RQDOG/(6PLWK Automated Fault Aalysis, IEEE Tutorial, IEEE PES, [2] M. Kezuovic, T. Popovic, Requiremets for Automated Fault Disturbace Data Aalysis, CIGRE Colloquium, SCB5, Sidey, Australia, Sept [3] M.H]XQRYLü < /LDR ³1HZ 0HWKRG IRU Classificatio Characterizatio of Voltage Sags, Electric Power Systems Research, Vol. 58, No. 1, pp , May 21, [4] 0.H]XQRYLü & 1DLO = 5HQ '5 6HYFLN - Lucey, W. Cook, E. Koch, Automated Circuit Breaker Moitorig, IEEE PES Summer Meetig, Chicago, July [5] 0.H]XQRYLü 6 /XR ' 5 6HYFLN ³$ 1RYHO Method for Trasmissio Network Fault Locatio Usig Geetic Algorithms Sparse Field Recordigs, IEEE PES Summer Meetig, Chicago, July [6] IEC 61850, Commuicatio etworks systems i substatios, i preparatio. [7] TLI, Ic.: DFR Assistat [Olie]. Available: [8] IEC , Commo format for trasiet data exchage (COMTRADE) for power systems, First Editio , IEC, [9] PTI. Ic. PSS/E, V#26, Software Maual, 1998