The traditional method of motor current signal characteristic analysis (MCSA) is commonly used in the fault diagnosis of rotor broken bar and eccentricity of asynchronous motor based on signal analysis. Because of low sampling frequency, strong base bourbon effect and other factors will lead to the drowning of characteristic frequency components, difficult to quantify the fault degree and other problems. Therefore, a fault diagnosis method of asynchronous motor based on adaptive particle swarm optimization sequential support vector machine (APSO-SSVM) is proposed. Firstly, empirical wavelet transform (EWT) is used to filter the original signal; then, the feature extraction of the filtered signal is carried out and input into the SSVM diagnosis model; finally, the APSO algorithm is used to determine the optimal hyperparameters of the SVM model in each order, so as to achieve accurate fault diagnosis of the number of broken rotor bars.