基于全息谱和支持向量机的水电机组振动故障诊断-水利水电工程专业论文.docxVIP

万方数据 万方数据 、 Abstract Title:VIBRANT FAULT DIAGNOSIS FOR HYDROELECTRIC GENERATING UNIT BASED ON HOLOSPECTRUM AND SUPPORT VECTOR MACHINE Major: Water Resource and Hydropower Engineering Name: Wei FANG Signature:_E W U Supervisor: Prof. Yuxia LI Signature:.，L;2Lμ似;乱 F Abstract There are a variety of vibration sources in hydroelectric generating 田让ts，and vibration becomes their common faults. In making the diagnosis of vibrant faults，vibration information containing frequencies and 缸nplitudes is used ，whereas phase information is always neglected. Meanwhile ，in intelligent diagnosis vibration signals obtained 丘om single monitoring point tend to be used as the sole signal so旧ce. They can indicate only partial vibration from one angle，but lack capability of revealing overall vibration of the whole UI由.To improve the reliability and accuracy of diagnostic results，it is essential to make full use of the amplitude，frequency and phase information，and conduct comprehensive and elaborate analysis of vibration signals from each monitoring point. Based on holospectrum and support vector machine ，a novel method is presented in this paper to intelligently diagnose vibrant faults of hydroelectric generating units. Firstly，given the intensive spectrum of the main shafts vibration signals，intensive spectrum correcting and zooming technology is used to carry out spectrum analysis，and the high-precision amplitude spectrum and phase spectrum 缸e therefore obtained. Secondly，through the 臼sion of amplitude information and phase information of vibration signals as well as the fusion of multi-sensor signals，two-dimensional holospectrum is drawn on the cross sections where vibration is monitored ，and fault features describing the shape and size of vibration ellipses in holospectrum 盯e then extracted 丘om it. At last，by using support vector machine ，mapping relationship is achieved between fault features and fault types in order to realize intelligent diagnosis. In simulation diagnosis，the propo