关键词: 永磁同步电机, 偏心故障, 经验模态分解, 希尔伯特变换, 分形维数, 遗传算法, 支持向量机

Abstract: Aiming at the diagnosis problem of permanent magnet synchronous motor (PMSM) dynamic eccentricity fault, a method of permanent magnet synchronous motor dynamic eccentricity fault diagnosis based on fractal box dimension and genetic algorithm-support vector machine (GA-SVM) was proposed. The empirical mode decomposition (EMD) was used to obtain the intrinsic mode function (IMF) of the vibration signal. The correlation coefficient method was used to filter out the effective IMF components. By performing the Hilbert transform on the effective IMF components, their instantaneous frequency and the fractal box dimension of instantaneous frequency as the feature vector were calculated. The GA-SVMwas used to identify the fault with the extracted features. The research results show that, compared with the traditional fault diagnosis method, the proposed method can achieve a higher-precision fault diagnosis of dynamic eccentric, which can reach 98.4%.

Key words: permanent magnet synchronous motor(PMSM), eccentric fault, empirical mode decomposition(EMD), Hilbert transform, fractal dimension, genetic algorithm(GA), support vector machine(SVM)