Abstract:  Aiming at the problem that when support vector machine (SVM) was applied to bearing fault classification, the traditional intelligent algorithm optimization of SVM parameters had problems such as slow optimization speed, more adjustment parameters, and easy to fall into local optimal values, a fault diagnosis method based on CEEMDAN and SSA-SVM was proposed. The fault feature extraction and SVM parameter optimization of rolling bearings were studied. A new swarm intelligence optimization algorithm sparrow search algorithm (SSA) was introduced to optimize the parameters of the SVM to improve the optimization speed and the accuracy of bearing fault classification. Firstly, the complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) algorithm was used to decompose the signal to obtain several intrinsic mode functions (IMF). Then the correlation coefficient method was used to select the useful IMF components and recombine them. Finally, the multi-scale entropy of the reconstructed signal was calculated as the feature vector and inputted into the SVM optimized by SSA for fault classification. The results indicate that this method can accurately obtain fault information and has high recognition accuracy. Comparing with SVM optimized by PSO and GA, this method has better fault diagnosis and classification performance.

Key words:  complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN); multi-scale entropy; sparrow search algorithm(SSA); support vector machine(SVM);fault diagnosis