摘要：

Engine knocking feature extraction using wavelet package transform is studied in this paper.Vibration signals acquired by an accelerometer were analyzed.The resonance frequency generated by knocking component is determined by the estimation of power spectrum density(PSD).An autoregressive(AR)model is made for the narrow data segment containing the knock energy,and the resonance frequency is estimated with Burg algorithm.The extent of wavelet packet decomposition is decided according to the result.The coefficients of sub-bands are chosen with a method proposed,so that knock feature is obtained,while noise is removed reasonably.Analysis in real vibration signals indicates that the proposed method is better than the conventional filtering method and basic wavelet analyzing method for the light knocking detection.

展开