Research on Motor Fault Diagnosis Based on Improved Support Vector Machine (SVM)
Xinyang Wang, Liting Hu, Mingjian Wei, Yu Liu, Xiangyu Niu · International Journal of High Speed Electronics and Systems · 2025
Motors have been widely used in the manufacturing industry. The stable operation of motors is the premise for ensuring the safety of human production and life. When a motor fails, it is necessary to diagnose it effectively. With the development of signal processing technology and artificial intelligence fault diagnosis, how to diagnose motor faults more quickly and efficiently has always been a research hotspot in the field of equipment diagnosis. The vibration signal of motor faults is difficult to decompose and difficult to extract features, and traditional machine learning has low accuracy in motor fault classification. In view of the difficulty in extracting motor fault features and the low accuracy of motor fault classification by traditional machine learning, we proposed a motor fault diagnosis method that combines ROA-VMD with wavelet packet information entropy features and Golden Jackal Optimization (GJO) to optimize SVM. The wavelet packet information entropy features of the signal after decomposition and reconstruction by ROA-VMD are used as the input of the SVM model, and GJO is used to optimize the selection of SVM kernel function parameters and penalty factors to improve the classification accuracy of SVM. Based on the simulation and comparative analysis of data provided by the laboratory of the University of the West of the United States, the GJO-SVM model has a higher classification accuracy for motor faults. The simulation results show that the GJO-SVM model reduces the number of iterations by 48 and 72 compared with the PSO-SVM and FOA-SVM models, and improves the fitness (classification accuracy) by 7.5% and 1.5%.