Study On The Nonlinear Characteristics Extraction Of Faulty Vibration Signals Generated By Gear Meshing

Gears are very common equipment in mechanical system. Gear fault will not only terribly damage the gear itself and normal work of the system, but also endanger the operators’ life safety, bringing about heavy economic losses and wide-ranging social influence. Therefore, it is very urgent to research the gear fault diagnosis.Vibration signals of various gear faults are collected on stimulant fault experiment tester. Advanced nonlinear signal processing methods are used in this work. Very useful fault features have been extracted using them. Consequently fault identification and diagnosis of various single fault and complex fault have been done. The result shows the efficiency of proposed methods.(1) The common invalidation forms and vibration cause of gears are introduced. At the same time the suitable scope and characteristics of common methods for diagnosing gears are analyzed in detail.(2) The process for analysis of vibration signals of gear faults is explained in detail. The EEMD and Wavelet are used to deal with the vibration.(3) A new nonlinear feature extraction method is presented. To analysze different failure modes, the manifold learning is used to adaptively investigate the fault characteristics, so as to eliminate the interference characteristics of the fault pattern recognition.(4) The SVM is adopted to perform the fault identification. Time SVM method is introduced into identify and diagnose fault. Fault features acquired by manifold learning are taken as inputs for SVM. The network testing result shows that the classification and diagnosis of various single faults have been done perfectly.The research is the development for theory and methods in gear diagnosis field. The methods are applied in actual fault diagnosis and good results are obtained. From engineering viewpoint, it has some important academic value.