关键词: 齿轮故障诊断, 多尺度, 时变滤波器, 稀疏分解, 线调频小波, 自适应

Abstract: A new method for adaptive time-varying filter design and its application in gearbox fault diagnosis is proposed. There are lots of components in gearbox vibration signals, especially for multi-stage gearbox, including numbers of components corresponding to meshing frequency, and the key point of gearbox fault diagnosis is the judgments about modulation of meshing frequencies. So, how to extract the AM-FM components from vibration signals is the vital point to be dealt with. The sparse signal decomposition method based on multi-scale chirplet can successfully extract components, whose frequencies are time-varying, from non-stationary signals, and is suitable for extraction of meshing frequency. The adaptive time-varying filter, whose center frequency and band width is set to be the meshing frequency extracted and multiple rotational frequencies respectively, can be designed and used in gearbox fault diagnosis. Each AM-FM signal can be filtered from non-stationary signals by the adaptive filter. The adaptive filter designed by using the proposed method can change the center frequency and band width according to the signals themselves. It keeps useful frequency components of signals and inhibits useless ones. Simulation and application examples prove the effectiveness of the method.

Key words: Adaptive, Chirplet, Gearbox fault diagnosis, Multi-scale, Sparse decomposition, Time-varying filter