Application of wavelet packet transform in the knock detection of gasoline engines

In order to extract knock feature for gasoline engines, vibration signals are acquired by an accelerometer mounted on the engine block head; a method based on wavelet packet transform is researched in this paper. Power spectrum density (PSD) estimation is used to determine the band range of the resonance frequency generated by knock component. An autoregressive (AR) model is made for the short data segment which contains knock energy, and the resonance frequency is estimated with Burg algorithm. The result decides which layer of wavelet packet decomposition is needed. Then the coefficients of sub-bands are chosen with a proposed rule, so that the knock feature is obtained while noise is reduced significantly. Real vibration signals analysis indicates an improvement in signal-to-noise ratio (SNR) with the proposed method; its performance is better than basic wavelet analysis method for light knock detection.