电气专业英文文献

1、An EXPert System for Tra nsformer Fault Diag no SiS US ing Dissolved GaSAn alysis1. INTRODUCTIONThe power tran SfOrmer is a major apparatus in a power system, and its COrreCt funCtiOning its vital to minimize SyStem outages, many devices have evolved to mon itor the Serviceability of power tran SfOr

2、mers. TheSe devices, SUCh as, BUChhOlZ relays or d if fere ntial relays, resp ond only to a SeVere power failure requiri ng immediate removal of the transformer from service, in WhiCh case, OUtageS are in evitable. Thus, PreVe ntive tech niq UeS for early detecti on faults to avoid OUtageS would be

3、valuable. In this way, an alysis of the mixture of the faulty gases dissolved in in SUIati on oil of power tran SfOrmer has received worldwide recog niti on as an effective method for the detecti on of on CiPie nt faults. Many researchers and electrical Utilities have reported on their experie nee a

4、nd developed in terpretative Criteria on the basis of DGA. However, Criteria tend to Vary from Utility to utility. Therefore, tranSfOrmer diag no SiS is still in the heuristic stage. For this reas on, kno Wledge-based PrOgram ming is a Suitable approach to impleme nt in SUCh a diag no StiC problem.B

5、aSed on the in terpretati on of DGA, a prototype of an expert SyStem for diag no SiS of SUSPeCted tranSfOrmer faults and their maintenanee PrOCedUreS is proposed. The Sig nifica nt SOUrCe in this kno Wledge base is the gas ratio method. Some limitati ons of this approach are overcome by incorporatin

6、g the diagnoStiC PrOCedllre and the Syn thetic expertise method. Furthermore, data bases adopted from TPC'S gas records of tranSfOrmerS are inCorporated into the expert SyStem to inCreaSe the PraCtiCal PerfOrma nee. Un Certa inty of diag no SiS is man aged by USing fuzzy Set con cepts. ThiS expe

7、rt SyStem is con StrUCted With rule based kno Wledge represe ntati on, SinCe it Can be expressed by experts. The expert SyStem building tool,knoWledge Engineering SyStem(KES), is USed in the development of the knoWledge SyStem because,it has excelle nt mach ine in terface that PrOVideS SUggeSti ons.

8、 Moreover,its inferenee Strategy is Similar to the MYCIN. A famous rule-based expert SyStem USed for medical diag no sis. The Un Certa inty of huma n qualitative diag noStiC expertise, e.g., key gas an alysis, and ano ther qua ntitative imprecisio n, SuCh as, norms threshold and gas ratio boun darie

9、s etc., are smoothed by appropriate fuzzy models. With the results of SUCh implementation, different Certainty factors will be assigned to the COrreSPOnding expertise Variables. Both event-driven(forward Chaining) and goal-driven (backward Chaining) inferenCeS are USed in the inference engine to imp

10、rove the in ference efficie ncy. To dem on St rate the feasibility of the proposed expert system, around hundreds of TPC historical gas records have been tested. It is found that more appropriate faulty types and maintenance SUggeStiO ns Can SUPPOrt the maintenance PerS on als to in CreaSe the PerfO

11、rma nee of tran SfOrmer diag no sis.2 DEVELOPMENT OF DIAGNOSIS AND INTERPRETATIONLike many diag no StiC problems, diag no SiS of oil-immersed power tran SfOrmer is a skilled task. A tran SfOrmer may fun ctio n well exter nally With moni tors, while some inCiPient deterioration may occur internally t

12、o CallSe a fatal problem in the latter developme nt. ACCOrd ing to a JaPa nese experie nee, n early 80% of all faults result from inCiPient deteriorations. Therefore, faults should be identified and avoided at the earliest POSSible Stage by some PrediCtiVe maintenance tech niq ue. DGA is one of the

13、most popular tech niq UeS for this problem Fault gases in tran SfOrmerS are gen erally PrOdUCed by oil degradation and other insulating material, e.g., cellulose and paper. TheOretiCally, if an in CiPie nt or active fault is PreSe nt, the in dividual dissolved gas COnCen trati on, gass ing rate, tot

14、al COmbUStible gas(TCG) and cellulose degradati on are all Sig nifica ntly in creased. By USing gas ChrOmatOgraPhy to an alyse the gas dissolved in a tran SfOrmer's in SUIati ng oil, it becomes feasible to judge the in CiPie nt fault types. ThiS StUdy is COnCerned With the following representati

15、ve COmbUStible gases; hydroge n( H2), metha ne(C2H2), etha ne(C2H6), ethyle ne(C2H2) and carbo n mon OXide(CO).Many interpretative methods based on DGA to the nature of inCiPient deterioration have been reported. EVen Under normal transformer Operational con diti ons, some of these gases may be form

16、ed in side. Thus, it is n ecessary to build COnCentration norms from a Sufficiently large SamPling to assess the Statistics. TPC in VeStigated gas data from power tran SfOrmerS to con StrllCt its criteria. The developed knoWledge base in this PaPer is partially based On these data. On the hand, DOrn

17、erburg developed a method to judge different faults by rati ng PairS of COnCen trati OnS of gases,e.g., CWH, GH4D3H4, With approximately equal solubility and fusion coefficients. ROgerS established mare COmPrehensive ratio codes to in terpret the thermal fault types With theoretical thermod yn amic

18、assessme nts. ThiS gas ratio method WaS PrOmiSing because it eliminated the effect of oil volume and Simplified the ChOiCe of Units. Moreover, it systematically Classified the diagnosis expertise in a table form. Table 1 displays the ratio method as proposed by Rogers. The dissolved gas may Vary Wit

19、h the nature and SeVerity of different faults. By an alyz ing the en ergy den Sity of faults, it's POSSible to dist in guish three basic fault PrOCeSSeS:oVerheating(pyrolysis), corona(partial dischatge) and arcing discharge. Corona and arcing arise from electrical faults, while OVerheat ing is a

20、 thermal fault. Both types of faults my lead to deterioration, while damage from OVerheating is typically less than that from electrical stress. In feet, different gas trends lead to differe nt faulty types, the key gas method is ide ntified. For example, large amounts of CH and H are PrOdUCed With

21、minor arcing fault 4 qua ntities of CH 2aid C2H2 may be a SymPtOm of an arcing fault.3. THE PROPOSED DIAGNOSTIC EXPERT SYSTEMThiS StUdy is aimed at developing a rule-based expert SyStem to PerfOrm tran SfOrmer diag no SiS Similar to a huma n expert. The details of SyStem PrOCeSS ing are described be

22、low.3.1 The Proposed Diag no StiC MethOdDiagnosis is a task that requires experience. It is UnWiSe to determine an approach from only a few investigations. Therefore, this StUdy USeS the Synthetic expertise method With the experieneed PrOCedllre to assist the popular gas ratio method and complete Pr

23、aCtiCal PerfOrma nee.3.1.1 EXPerie need Diag no StiC PrOCedUreThe overall PrOCedUre of routi ne maintenance for tran SfOrmerS is listed. The core of this PrOCedUre is based on the impleme ntati on of the DGA tech niq ue. The gas ratio method is the Sig nifica nt kno Wledge source. Some OPerati on al

24、 limitatiO ns of the gas ratio method exist. The ratio table is Un able to cover all POSSible cases.Mi nimum levels Of gases must be PreSent. The solid inSUIatiOn involving Co and Co are han died Separately and the gas ratio codes have bee n developed mainly from a free - breathi ng tran SfOrmer. Ot

25、her diag no StiC expertise should be USed to assist this method. Norms, Synthetic expertise method and data base records have been in corporated to complete these limitati ons. The first SteP of this diag no StiC PrOCedllre beg ins by ask ing DGA for an oil SamPle to be tested. More importa nt relev

26、a nt information about the tranSfOrmer's condition, SUCh as the voltage level, the PreSerVatiVe type, the orvli ne-tap-cha nger(OLTC) state, the OPerati ng Period and degassed time must be known for further inference. NOrmS(Criteria) Set UP by TPC power tran SfOrmers' gas Characteristic data

27、 are the n USed to judge the tran SfOrmers* condition. For the abnormal cases, the gas ratio method is USed to diagnose tran SfOrmer fault type .If differe nt or Un knOWn diag no SiS results are found from these ratio methods, a further Syn thetic expertise method is adopted. After these procedures,

28、 different SeVerity degrees are assigned to allow appropriate COrreSPOnding maintenanee SUggeStiO ns.3.1.2 Syn thetic EXPertiSe MethOdThe ratio trend, norms threshold, key gas analysis and some expertise are con Sidered as differe nt evide nces to con firm some SPeCial fault types .In other words, m

29、ore Significant evidences have been collected for some SPeCial fault type, better assessme nt of the tran SfOrmer StatUS is obta in ed.The ratio trend Can be See n as a modificati on of the COnVen ti onal gas ratio and key gas method.ObViOUsly, the above gas trends should be incorporated With other

30、evidences Un der the experie need PrOCedllre for PraCtiCal use. Norms threshold, the gass ing rate, the qua ntity of total COmbUStible gas(TCG), the TPC maintenance expertise and the fuzzy Set assig nment are all importa nt evide nces con Sidered in the Syn thetic diag no sis.Other expertise based o

31、n a tran SfOrmer historical data base is also USed to an alyse the Characteristics of a CaSe tran SfOrmer. SeCti on 3.4 gives some details of these rules.3.2 EXPert SyStem StrllCtUreThe proposed diagnoStiC expert SyStem is composed Of COmPOnents, WOrking memory, a kno Wledge base, an inference engin

32、e and a man-mach ine in terface. Worki ng memory (global data base) COntainS the CUrre nt data releva nt to solve the PreSe nt problem .In this study, most of the diag no StiC Variables StOred in the data base are CUrre nt gas COnCen trati on, some are from the user, OtherS are retrieved from the tr

33、an SfOrmer's historical data base. Note that the fuzzy Set con CePt is in Corporated to Create fuzzy Variables on the request of SyStem reas oning. A kno Wledge re I at i on ship, WhiCh USeS these facts, as the basis for decisi on maki ng. The PrOdUCt ion rule USed in this SyStem is expressed in

34、 IF-THEN forms. A SUCCeSSfUl expert SyStem depe nds on a high quality kno Wledge base. For this tran SfOrmer diag no StiC system, the kno Wledge base in corporates some popular in terpretative methods of DGA, Syn thetic expertise method and heuristic maintenance rules. SeCtiOn 3.4 will describe this

35、 knowledge base. Ano ther SPeCial con SideratiO n in the expert SyStem is its inferenee engine. The inference engine controls the Strategies of reasOningand SearChing for appropriate kno Wledge. The reas oning Strategy employs both forward Cha inin g(data-drive n) and backward Cha inin g(goal-drive

36、n). FUZZy rules, norms rules, gas ratio rules, Syn thetic expertise rules and some of the maintenance rules and some maintenanee rules, USe forward Cha iningAS for the SearCh ing Strategy in KES, the depth first SearCh ing and ShOrt- CirCUit evaluati on are adopted. The former Can improve the SearCh

37、 efficie ncy by PrOPerly arranging the loCation of Significant rules in the inferenee procedures. The latter Strategy only SearCheS the key conditional Statements in the antecedent that are responsible for establishing Whether the entire rule is true or false. Taking the advantagesof these two appro

38、achesin the building and Structuring of a knoWledge base improves inference efficie ncy Sig nifica ntly.AS for man-machi ne in terface. KES has an effective in terface WhiCh is better than typical kno Wledge PrOgram ming Ian guages, SUCh as, PRoLoG or LISP. With the help of this in terface, the Capa

39、bility of trac ing, explai ning and trai ning in an expertSyStem is greatly simplified.4. IMPLEMENTATION OF THE PROPOSED EXPERT SYSTEMAn expert SyStem is developed based On the proposed in terpretative rules and diag no StiC PrOCedllreS of the overall system. To dem on Strate the feasibility of this

40、 expert SyStem in diag no sis, the gas data SUPPOrted by MTL of TPC have bee n tested. In TaiWan, the MTL of TPC PerfOrmS the DGA and SendS the results to all acting divisions re I at i ng to power tran SfOrmers. In return, these acti ng divisions are requested to collect and SUPPIy their tran SfOrm

41、er oil SamPleS Periodically.After analysing oil SamPles, more than ten years' WOrthy gas records are collected and Classified into three voltage level, 69KV, 16KV and 345KV . Thus, gas records for Onetran SfOrmer are composed of SeVeral groups of data .In the PrOCeSS of DGA interpretation, all o

42、f these data may be considered, but only the recent data WhiCh have Sig ni fica nt effects on diag no SiS are listed in the later dem on Strati on. In MTL, all gas COnCen trati ons are expressed by Pm in volume COnCen tratio n. 100 Pm is equal to 0.01 ml(gas/100ml(oil).From the expertise of diag no

43、sis, the no rmal State Can be con finned only by in SPeCti on of the tran SfOrmer*s norms level. I n practice, most of the tran SfOrmer oil SamPleS are no rmal, and this Can be in ferred SUCCeSSfully on the early executi on of this expert system. However, the SUCCeSS of an expert SyStem is main ly d

44、epe ndent on the Capability of diagnosis for the transformers in question. In the implementation, many gas records WhiCh are in abno rmal con diti on are chose n to test the JUStifiCati on of this diag no StiC system. A total of 101 tran SfOrmer records have bee n executed and the results are SUmmar

45、iZed in Table 5. Among those impleme nted, three are listed and dem on Strated.Show n in Table 5 are the results of 101 Un its of tran SfOrmerS in three types of remedy: normal, thermal fault and arc fault. After COmParing them With the actual State and expert judgeme nt, a SUmmary of results WaS ob

46、ta in ed. AS previously stated, one Unit of tran SfOrmer may in elude many groups of gas data. In evaluation, We depicted some key groups in one Unit to justify because some tran SfOrmerS may have different inCiPient faults during different OPerational stages. Some mistakes implemented from testing are CaUSed by the remaining oil in the oil SamPling container, Un Stable gas Characteristics of the new degass ing SamPle and some ObSCUre gas types .If more