开关磁阻电动机的故障诊断与容错控制研究

《开关磁阻电动机的故障诊断与容错控制研究》由会员上传分享，免费在线阅读，更多相关内容在工程资料-天天文库。

1、西藏工业大学破主学位论文摘要摘要开凳磁阻电桃駆动鑿统（g蜥紛融Re1uctanceDrive，越鸯特麟碱镰穂儷特醐^型电橇疆动券统。它兼骞髯步官旗橇变鑒调遮悉绕秘囊溅殼勰撬谖漢系统診餐秦，澜谶穗潮宽，调逊性能优診，盥具有较好舗弊错性。这些优点使得开关磁阻电机在家川电器、通用工业、航窝工业、伺服妤澜逮系统、牵攀1电梳、廉转速魄辘等方蕊褥夔1广泛应嚴。为了辘裔开关磁鑒锻晚系统静掰靠壤及稳定瞧，零变静对器蓑磁阻锂瓣系统静电极奉体上定予掰发生醵主要鼓黔定予一穗缝缀开路、一樱绕缀短路送秽婿究探讨。逢文奔缮了籀美磁鑒魄霉£累壤診基本络褥特患娃及敬簿渗糖憊必饗壤，莽葡谍艇

2、的研究现状邀杼了阐述。分析了开关磁隘电机故障诊断控术的研究裁撩现状以及忤惫瓣窥熬叙缴A工棒缝弼络铸智態放障诊麟技零，势对器关磁阻邀粮瓣常霓赦漳釉粪爱箕发生枫理澀行T掺辑o蜡究势建立了释蓑磁隧毫甑勵线性辩簿线靛数学構凝；建壹系绕蹤象鹅熬学模型是进杼各项研究工俸如控制壤储应硝、故龄诊断等酌鑒础。弁绍了箕数学模型建立方楼的硬究现状积存在闻黔，建立了定性静析的线性横燮，在对其内部蔑杂的电磁燕瀑进行了漾入分橱靛蕊躺主建照7零盛靛菲绫熬壤登，雾绘澄游粪及分析结论。舟缁了在故簿诊蔽方两广泛旋闱的神经网络散障诊断技术，萍饕重说鯛了率谖惩采麓匏诊舔骞法秘神糕殛络。狠摄采集豹典

3、篷故障样零集数摇梅造曼耀醋弹投褥貉绻鞫，辩努利聚掰構经飕绦缱菰猴静簿滚、口逮疯掌露邃率算法以及共轨梯魔簿法等聪网络进行7”｛J11练。然后用测试撵零集对嘲络避行7测试诊断以及轉法鳗I势辑比较。绦暴证明，媳翌罄齬燃终爨毒壁驾熬诊攀磁嘉，其垮煞共辘糖度葵法是毙较爨蜜的冀洼。最嚣辩开关磁鬻魄枫的绕缝廣阑短路赦障的靛生枫理擞敢障分析释了初步的研究少探讨。关键譎：嚣关磁隧瞧枫敛朦渗簸｛｝线慷模裁辩经阀缭8P算法舞黔蘸障筑嬉毂簿AbstractSwitchedReluctanceDriveisakindofnewelectricaidrivet5ha,thasthe.s

4、.pecial】,..T_doub1e•sa1lentpolestructurecharacteristics.1thasexcellenceofbothasynchronous,motorandd1rectcurrenlmotordrivesystems^thatisthewidesBxceffeSfper^ormance.TI1isvirtuemakesitsextensiveapp1icafioninhomeelectricappliances,universalindustry,aviationindustrlyiptsoerravnod由ridH丁

5、piWeidrmg)stsisn,tractionInoraertoimprovethere1iabi1ityandstabi1ityofthesyste〔in.Jhelhe^.isresearc〔・人i儿丨儿〔〔thefaultdiagnos1softhedrivesxstem,andtakestheonephaseshortandorlephasenopenfau1tasresearch®bljeetd)(tsisintroducedthebasicstructurecharacteristicsoftheswitchedreluctancedrives

6、ystem,thenecessityandthepresentresearchstatusoffaul[diagnosis.Analyzedthedevelopmentoftheswitchedreluctanced百©NcPdfuac比IdE4d鹉pHuOligentdiagnosistechnology,fr^aansa^i&95(?』%fiuu1toftheSRDandgenerantmechanismofthefault.Studiedandestablishedthemathematicmode1oftheSRD.Modelihecontro1st

7、rategyandttusandtheprob1emsofmode1heSRD.AfterdeeplyanalyeSRD,establishedthenonngiSthebasicoftheotherresearchworkssuchaStfiftaruo^d8?fi1psress*entresearchstaingSRD.thenestab1ishedthe1inearmodeloftzedtheinner.(.electromagnet1smrelat1onofth上cli°^aarrd叫aF*mPjShfeandanalysia1eikpli(co企

8、1ntro'ducedtheneuralnetwor