Fault diagnosis for power unit based on wavelet packet PCA-SVM

In this paper, a new method of fault diagnosis for power unit based on wavelet packet PCA-SVM is proposed. Firstly, using wavelet packet transformation to extract each band of energy as the initial samples; Secondly, taking principal component analysis to excavate the features of the initial samples, eliminating the correlation between data while ensure the integrity of data as far as possible, then the smallest diagnostic features were got. The fault diagnosis model based on SVM and the smallest features can effectively reduce the computational complexity and the difficulty of obtaining fault characteristics. Simulation results show that the proposed method can effectively shorten the time of diagnosis, improve the diagnostic efficiency, this method is an effective way to diagnosis the fault for power unit.