A step-by-step parameter-adaptive FMD method and its application in fault diagnosis

Abstract A newly proposed method, feature mode decomposition (FMD), can effectively enhance signal features while decomposing the signal. This feature is beneficial for analyzing weak vibration signals. However, the input parameters (the segment number K , the filter length L , and the mode number n ,) significantly influence the decomposition performance and efficiency. Based on the analysis of filter properties and decomposition performance of the FMD method, a step-by-step parameter-adaptive FMD method is proposed. First, parameters K and L are optimized; second, the parameter n is determined. In addition, a comprehensive evaluation indicator, the ratio of sample entropy and ensemble kurtosis, is constructed considering both the periodic impact characteristics of fault signals and the noise intensity to create objective functions for each step. Compared with the methods of variational mode decomposition spectral kurtosis method and the wavelet packet decomposition, the proposed method exhibits better decomposition performance: the amplitude has increased by nearly ten times for the simulation data and six times for the actual engineering data, and three evaluation factors (the crest factor, the impulse factor, and the kurtosis) have higher value. Therefore, it can be concluded that the proposed method has better superiority in identifying weak periodic fault features.