Based On The Hht And Support Vector Machine (svm) Of Rotating Machinery Fault Diagnosis Research

Hilbert-Huang transformation, as a new kind of signal processing method, can effectively extract characteristic information of signal, which has extensive application prospect in fault diagnosis of rotating machinery.This paper studies the endpoint problems existing in the EMD. According to the features of signal waveform at both ends can’t be isolated existence, full considering the change of the signal itself, describing the mean half wave continuation method. At the same time, in view of the transient process is short, the fault happening in moment, can get less data, this paper introduces a kind of endpoint continuation method based on the LS-SVM regression.After solving the endpoint effect, using Hilbert-Huang transformation to extract the feature information of rotating machinery fault, and thus classify and monitor the fault types. After EMD decomposing the fault data, considering using the fault feature extraction based on energy firstly, taking all the energy of the IMF component as characteristic vector, and through the LS-SVM classifier to classify types. However, the more IMF component can produce the more characteristic vector, then it will occupy much time in the classification, go against the real-time of realizing fault diagnosis, therefore this paper describes an information separation method based on the difference degree. Use this method to select the IMF component of fault information which has relatively concentrated, and then classify the faults. Finally, this paper introduced the fault feature extraction method based on the natural modal functions envelope spectrum. And through the method to analysis the measured signal of the running state of rotating machinery in four cases, and extracting the fault feature information accurately, reflecting the state of rotating machine precisely.The research results in this paper show that, the fault diagnosis method based on the Hilbert-Huang transformation and LS-SVM rotating machinery can make good evaluation of the operation status of the rotating system, classify the fault type accurately,and be worth to extensive application.