<FONT face=Verdana>This paper presents an adaptive signal decomposition method. The core of the proposed method is to decompose the observed signal into a series expansions of waveforms which best matches the signal local structures. These waveforms are calculated by the nonparametric waveform estimation method. With the proposed method

the adaptive adjustment of template signals makes it unnecessary for the method to require priori information

so that it has better flexibility and adaptability in practical applications. The performance of the proposed method is confirmed by testing its adaptability and convergence. The extraction results of simulation signals are compared with which of the feature extraction method based on the traditional MP decomposition and EMD decomposition

thus verified the feasibility and effectiveness of the proposed method farther.