A Fault Diagnosis Method for AC Contactor Based on Parametric Optimization Feature Mode Decomposition

The actual working conditions of AC contactors are complex and severe, and for the problem that it is difficult to effectively extract the mechanical fault information in its vibration signals, this paper proposes a fault diagnosis method based on parameter-optimized feature modal decomposition. Firstly, considering that the original feature modal decomposition (FMD) does not have the defect of adaptivity, it needs a priori knowledge to set the core parameters, and proposes a grid search method based on the ratio of squared envelope entropy and squared envelope spectral Gini coefficient (SEE-SESGI) to determine the number of modes and filter length of the FMD automatically, and reconstructs the signals decomposed by parameter-optimized FMD based on correlation coefficients between the components and the original signals. reconstruction is performed. Secondly, the composite multiscale sample entropy (CMSE) of the reconstructed signals is calculated as the feature data, and the feature matrix is downscaled using principal component analysis (PCA) and t-distribution-stochastic neighborhood embedding algorithm (t-SNE). Finally, the experimental signals are collected and recognized and diagnosed by using four classification learning models, and compared with the recognition accuracy and other performances without parameter optimization, which show that the proposed method has higher accuracy and can effectively diagnose the faults of AC contactors.