Feature extraction of an engine vibration signal based on dual-tree wavelet package transformation

In order to extract the relative energy feature of engine cylinder head vibration signals,a new method based on dual-tree wavelet package transformation and adaptive block threshold de-noising was proposed.The dual-tree complex wavelet package transformation(DT-CWPT)made sure that the coefficients decomposed by tree a and tree b could be complementary information at each scale,so the DT-CWPT could achieve approximate shift invariance and decrease the information missing.The adaptive block threshold de-noising method gained the best threshold λ and the optimal neighborhood effect length L for signal de-noising and could obtain higher signal-to-noise ratio.Also the causes of frequency aliasing and frequency bands derangement were discussed and Fourier transformation method was applied to suppress the false frequency in the dual-tree wavelet package transformation.The simulated signal and the measured signal test results showed that the proposed method has a good de-noising performance and it is more effective in fault feature extraction.