Vibration subsequence alignment method based on adaptive variational modal extraction

Abstract During the start-stop process of rotating machinery, there will be many switching conditions. When we sample a long vibration sequence with equal length, it is easy to cause the problem that the starting point position (initial phase) of the vibration subsequence is different, which leads to the inaccurate alignment of the sequence. And because there are many harmonic components and noises in the vibration subsequence, the effect is poor when the vibration subsequence is directly used for sequence alignment. Variational Modal Extraction (VME) can extract specific modes without knowing the number of modes in the signal. In this paper, the initial center frequency is determined by using short-time Fourier transform (STFT), and the penalty factor is determined by mapping relationship, so that VME can adaptively obtain specific modal components in vibration subsequences under different working conditions. Then the initial phase value is calculated by curve fitting of modal components, and the sequence is cut and aligned according to the phase difference, so that the multi-segment vibration subsequence can be aligned well. In this paper, a new alignment method of vibration subsequences under various working conditions is proposed. Compared with the current methods, the alignment effect of the proposed method is better.