基于大型旋转机组的非平衡状态小波模式识别与神经网络预测技术研究-机械电子工程专业论文.docxVIP

ABSTRACTABSTRACT ABSTRACT ABSTRACT Electromech卸ical equipment is vcry imponant in the proccSs of m锄ufacturc． ne佗forc the pro铲ess of fault diagIlosis柚d trend prcdicti伽tcchnoIo百es will not only p砌mote the d“elopment of sciencc趾d tcchnology but also haVe c九Icial social ande咖omicsi印诟cancc．Elect咖nechanicaIequipment∞nline缸锄dnostationary fault diagIIosis锄d intelligcnt仃end predidion technolo舀es盯e Very import孔t for equjpment condition mainten柚cc． C0mbining tlle requests Of the elcctmmcchanical system nOnline缸卸d∞ stationary fault dia舶osis aIld trend prediction with me anificial intelligent tcchnology developmcnt，the imelligent fault dia卸osis and仃cnd predicti伽teclInolo醇es based 伽knowledge arc studied by using wavelct孤d wavelet packet柚alysis combining NN technolo酉es．1nhe main∞ntributions of this dissenation缸e驯mmarized as follows： FifstIy，hl view Of large·scale fcvOlves the mechanical si印al non—steady charaderistic，may simultaneously carry on the time domain and t量le仃cqu锄cy r柚ge analySis charaderistic using tlle wavelet仃ansfo姗ation to tlle si印al，rcalizes the si印al strangc point the examination，disappears chifp，processing锄d so∞trcnd analysis．Thjrd methods a陀in协础uced for de—noisc，the affecti∞of de-nOise result is analysed with different choicc of Daubechies wavelet orders． Secondly’withdmwing the characteristic value of Vibrate the signal of lhe rotary machines by using wavelet packet technoIogy，Mo questiolIs is proposed蛆d solVed in the pmceSs Of the characteristic vaIue withdl毽wing，one question is the choicc of sampling ffequency，which may be solved through the interpolation metllod，the Other is that according to the studies gOal to dctemline wavelet primafy fIlnction；Studied in this foundation iS tlle method of withdrawing the partial charactcriStic Value with wavelct packet；Finall y，fuzzy卸d the ncn，e ne伽ork thcory盯e gathered together and a kind of theor)r proposed is wavelct packetfllzzily clustering diagnosis net、ⅣOrk，fivc groups Of ail blowers sigIlals which was gatherc