A Fault Feature Extraction Method for Rolling Bearings Based on Refined Composite Multi-Scale Amplitude-Aware Permutation Entropy

Aiming at the problems of unclear early fault characteristics and difficult extraction of rolling bearings, a new nonlinear dynamic analysis method called refined composite multi-scale amplitude-aware permutation entropy (RCMAAPE) is introduced in this paper. Firstly, RCMAAPE is used to extract features from the bearing life data, and Chebyshev's inequality is used to establish a health threshold to evaluate the performance degradation state. Secondly, RCMAAPE is also used for bearing fault diagnosis. Both experimental results prove that RCMAAPE could extract fault characteristics effectively. RCMAAPE can accurately reflect the degradation trend of bearing in the whole life process, and is especially sensitive to the early failure of the bearing. RCMAAPE is also able to effectively identify the states of bearing faults. Especially after selecting the features, RCMAAPE only needs a small number of features to effectively identify the different states of the bearing and the recognition accuracy is up to 100%. Compared with the existing methods, the proposed method can extract fault features more effectively, has higher computational efficiency and obvious advantages.