資源描述:
《變頻調(diào)速系統(tǒng)的故障診斷研究》由會員上傳分享,免費(fèi)在線閱讀,更多相關(guān)內(nèi)容在工程資料-天天文庫。
1、摘要木文結(jié)合國內(nèi)大型鋼鐵行業(yè)的實際情況,概述了調(diào)速系統(tǒng)的組成與結(jié)構(gòu),著重分析了交■交變頻器在同步電動機(jī)調(diào)速系統(tǒng)中的工作方式,并介紹了交■交變頻器的常見故障??紤]實際生產(chǎn)屮的交■交變頻器英內(nèi)部信息難以獲取,選擇利用相關(guān)輸入/輸出電量(如電壓、電流等)作為檢測對象,來診斷變頻器內(nèi)部故障。針對實際變頻器利用實驗實際獲取數(shù)據(jù)難的情況,對具體工作狀況,應(yīng)用MATLAB軟件對整個調(diào)速系統(tǒng)進(jìn)行仿真實驗。在仿真模型的基礎(chǔ)上,模擬了在實際工況中常見故障,如晶閘管開路、短路的情況,并獲取三相輸出電流的試驗數(shù)據(jù),通過FFT變換,將變換后的數(shù)據(jù)作為診斷的基本信息。通過對國內(nèi)外多
2、種神經(jīng)網(wǎng)絡(luò)和智能故障診斷技術(shù)的分析比較,并考慮到直接檢測變頻器內(nèi)部冇關(guān)信息的難度,本文提出并研究了基于神經(jīng)網(wǎng)絡(luò)的變頻器故障診斷技術(shù),以變頻器輸入/輸出有關(guān)電量(電壓、電流等)特征值作為神經(jīng)網(wǎng)絡(luò)的輸入,通過神經(jīng)網(wǎng)絡(luò)的自學(xué)習(xí)方式,對變頻器內(nèi)部品閘管故障進(jìn)行定位。并針對BP神經(jīng)網(wǎng)絡(luò)的不足,引進(jìn)了帶偏羞的遞歸神經(jīng)網(wǎng)絡(luò)來改進(jìn)故障診斷的有效性。通過對比分析可知,大大改進(jìn)的神經(jīng)網(wǎng)絡(luò)擬合效果和泛化能力強(qiáng)。仿真實驗和研究表明,該方法能迅速的進(jìn)行故障判斷,同吋對故障發(fā)生的晶閘管故障準(zhǔn)確定位,有效的減少了排查故障的時間,提高了交■交變頻器運(yùn)行的可靠性,確保了生產(chǎn)正常進(jìn)行,為交
3、■交變頻器的智能故障診斷技術(shù)應(yīng)用捉供了新的思路。關(guān)鍵詞:交?交變頻器仿真故障診斷AbstractThecharacterofCycloconverteriswithoutDClink,smallvolumeandsoon.Thencycloconverteriswidelyusedintheconditionoflowrotatespeedandhighcapacity,asifsusedinthedrivesystemofrollingmill.Duetotheimportantfunctionofcycloconverterinthesystem,o
4、ncethefaultofcycloconvertehappens,ifthepositionandreasonofthefaultcan'tbefoundintime,itwillinducegreateconomicloss.Atpresenttheintelligentfaultdiagnosisforfrequencyconverterisnotperfectlyachieved,andtheresearchesonAC-DC-ACconverterfaultdiagnosisaremore,thentheresearchisveryessent
5、ialandpracticaltosetupthecycloconverterfaultdiagnosissystem.Thethesisassociatestheactualconditionofsteelindustryincountry,summarizesthecomposingandframeofthesystem,emphasizestheanalysisofworkmethodofcycloconverterinsychchro-motorspeedgoverningsystem,andintroducesthecommoncyclocon
6、verterfaults.Aimingattheactualworkcondition,thespeedgoverningsystemissimulatedbyMATLAB.Basedonthesimulationmodel,thecommonfoultissimulated,forexampleSCRshortoropencondition,thentheexperimentdataofthreephaseoutputcurrentisgotandtransformedbyFFT?Thetransfonneddataistakenasthebasicf
7、aultdiagnosisinfbrmation.BystudyingandanalyzingthevariousNeuralNetworkandintelligentfaultdiagnosis,associatingthedifficultywhichistomeasuretheinteriorsignalofcycloconverter,thethesisputsforwardthecycloconverterfaultdiagnosiswhichbasesonNeuralNetworkandtakescycloconverterinputvolt
8、ageandoutputcurrentastheinputofNN.Accord