Abstract: In view of the limitation of mass calculation,slow analysis velocity and complicated feature extraction of traditional time-frequency,the feature extraction method of diesel engine based on time-frequency distribution singular value spectrum was proposed.The discrete Dual-Tree Complex Wavelet Packet Transform (DT-CWPT) transform method and the discrete time-frequency distribution was studied,then the feature was extracted by singular value decomposition.The comparison of calculation efficiency and feature extraction effect was analyzed based on different working condition test of diesel engine.The results indicate that the method has high computation efficiency and good performance of accuracy.

Key words: dual-tree complex wavelet packet transform, discrete time frequency distribution, feature extraction, time frequency singular value spectrum