A Fractional Steepest Ascent Morlet Wavelet Transform-Based Transient Fault Diagnosis Method for Traction Drive Control System
Chao Yang, Zhiliang Wu, Tao Peng, Hongqiu Zhu, Chunhua Yang
Abstract
In this article, a fractional steepest ascent Morlet wavelet transform (FSAMWT) method with parameters optimization for transient fault (TF) diagnosis in traction drive control system (TDCS) of high-speed trains is proposed. Compared with the existing literature focusing on the TF with distinct characteristics, such as obvious amplitude changes and specific frequency components, the proposed method can deal with some TF scenarios in TDCS, which characterize with small amplitude, short duration, and energy feature pattern aliasing with noise. To be specific, the FSAMWT is first proposed. Then, a cost function of obtaining the minimum Tsallis entropy of FSAMWT is proposed to solve the parameters optimization of Morlet wavelet. Subsequently, the energy distribution of the transient kurtosis of FSAMWT coefficients is used to overcome the limitation of singular value decomposition-based methods, in which the demarcation point in singular values between the transient signals and the noise are inconspicuous. Finally, the proposed FSAMWT-based TF diagnosis method is used to diagnose six common TFs in TDCS, and the hardware-in-the-loop-based experiments and their result analysis shows the effectiveness and merits of the proposed methods.