Bearing fault diagnosis of in-wheel motor based on SMOTE-IGWO-RF with small sample
The complex and dynamic operating environment of in-wheel motors may lead to bearing failures,posing risks to the safety of electric vehicle operation.In order to address the low identification accuracy of traditional fault diagnosis methods under small sample conditions,a bearing fault diagnosis method for in-wheel motors based on SMOTE-IGWO-RF is proposed.Firstly,the training dataset is expanded using synthetic minority over-sampling technique(SMOTE)to generate fault samples similar to the real sample distribution,and principal component analysis(PCA)is used to optimize their time and frequency domain features.Then,by introducing a nonlinear convergence factor and Levy flight strategy to improve the traditional grey wolf optimization(GWO)algorithm,the parameters of the random forest(RF)model are optimized using the improved grey wolf optimization(IGWO)algorithm.Finally,the bearing fault diagnosis model for in-wheel motors based on SMOTE-IGWO-RF is implemented to identify fault states,and experimental verification is conducted on the in-wheel motor test bench.The results indicate that the proposed bearing fault diagnosis method for in-wheel motors achieves an average accuracy exceeding 96%under seven different speed conditions,demonstrating high precision and stability.Compared to genetic algorithm(GA),particle swarm optimization algorithm(PSO),and GWO optimized RF,the proposed IGWO-RF model achieves diagnostic accuracies exceeding 90%under three small sample training sets.Moreover,its accuracy is significantly higher than the other three comparative algorithms,effectively realizing bearing fault diagnosis of in-wheel motors under small sample conditions.
NETL
NSTL
万方数据
