Multicomponent Nonstationary Fault Feature Extraction Based On Match Chirplet Transform
Jingbo Liu, Zong Meng, Zhaolin Cui, Shuhan Quan, Jing Liu · 2023
Mechanical fault signals are usually multicomponent and non-stationary, which makes the accurate extraction of multicomponent fault features a challenging task. Therefore, this paper proposes an Match Chirplet transform that can adaptively match the time-varying laws of the components to handle multicomponent nonstationary signals. It estimates the Chirp rate parameter in advance from a rough time-frequency result. Then the basis which match the signal is constructed for secondary time-frequency transformation to obtain the optimal time-frequency representation. Finally, post-processing is performed along frequency direction to extract the point with the maximum local energy. Excellent energy concentration and noise robustness were verified through a multicomponent simulated signal. Furthermore, the performance of the proposed algorithm was tested in the actual signal section by applying it to a bat echo signal and bearing fault signals.