Application of HHT in SRM fault feature extraction

Switched Reluctance Machine (SRM) has magnetic field with strong saturation nonlinearity features, complex mathematical models and its fault output is mainly unsteady signals of strong coupling multi-physical field, which easily floods effective fault characteristics and make it difficult to extract. In this paper, Hilbert-Huang transform (HHT) is introduced to SRM fault feature extraction method to solve the problems aforesaid. Firstly, Empirical Mode Decomposition(EMD) is utilized to decompose the bus current of the faulted motor into several simple Intrinsic Mode Function(IMF) to resolve the problem of unsteady characteristics of complex fault signals. Secondly, primary IMF components are selected to form the matrix of initial parameters to calculate both the energy of singular values and the parameters of energy entropy of the matrix, which is used as a feature vector. Finally, multi-classifier based on support vector machine (SVM) are used to identify the extracted small-sample fault feature vector for classification. After verification by simulation, this method can effectively reduce the complexity of the fault signals, redundant data of faults and increase the accuracy rate of fault identification. Its application in SRM fault diagnosis has theoretical and practical value.