Aiming at the difficulty of feature extraction and low recognition accuracy of rolling bearing fault signals,an improved gray wolf optimization algorithm based on dimensional learning(DIGWO)was proposed to optimize variational mode decomposition(VMD)and compound multi-scale permutation entropy(CMPE)bearing fault diagnosis method.Firstly,the grey wolf optimization algorithm(GWO)was modified into DIGWO,cosine convergence factor a and individual wolf ω position updating methods,and multiple intrinsic mode functions(IMFs)were obtained by using the adaptive optimization of VMD decomposition of DIGWO algorithm.Then,the eigenvalues of IMFs were calculated by compound multi-scale permutation entropy,and the eigenvectors of fault were constructed by selecting the features of appropriate dimension.Finally,DIGWO algorithm was used to optimize the penalty coefficient C and radial basis function g of support vector machine(SVM),and a DIGWO-SVM rolling bearing fault diagnosis classifier was established.The research results show that the CMPE-based DIGWO-SVM rolling bearing fault diagnosis method can effectively identify the running condition of bearings,and the recognition accuracy is 99.42%,which is 7.75%and 1.68%higher than PSO-SVM and SSA-SVM methods,proving that the classification performance of this method is more advantageous in the rolling bearing fault diagnosis.
维普
万方数据