Extraction method for signal effective component based on extreme-point symmetric mode decomposition and Kullback–Leibler divergence

Abstract

Data processing is widely used to extract effective component from original signal, which is essential in mechanical condition monitoring and fault diagnosis. In order to solve the invalid component and non-stationary feature in the measured signal, the extraction method for effective signal component is proposed based on extreme-point symmetric mode decomposition (ESMD) and Kullback–Leibler (K–L) divergence. This method fully integrates the characteristics of ESMD in self-adaptive decomposition and the advantages of K–L divergence in measuring the distance between different signals. The effective and invalid components of non-stationary signal are automatically separated by ESMD, and the effective components are further identified through K–L divergence calculation. Some analyses of simulated data and experimental data were investigated. And the effect of the proposed method in effective component extraction was emphatically explored. Research results indicate that the proposed method can adaptively acquire effective signal components with higher accuracy. Moreover, compared with the classic method, it is more efficient in the extraction of effective components from complex signal. In addition, this research solves the interference problem of invalid signals and accurately reconstructs the desired useful signal.