Fault Diagnosis of Rolling Bearing Based on VMD and POA-SVM
To address the problem of fault state identification of three-phase motor bearings,a fault identifi-cation method combining Variational Mode Decomposition(VMD)and Peafowl Optimization Algorithm(POA)to optimize Support Vector Machines(SVM)was proposed.The different fault signal data of bear-ings were decomposed into multiple Intrinsic Mode Function(IMF)by VMD.First,according to the sample entropy,the appropriate IMF reconstructed the optimal feature signal;Secondly,the time domain features,energy entropy and multi-scale sample entropy of the feature signal were calculated to form a multi-dimen-sional feature vector matrix.Finally,the peacock optimization algorithm was used to optimize the penalty pa-rameters and kernel function of SVM,and the POA-SVM diagnosis model was established,and the con-structed multidimensional eigenvector matrix was input into the model for diagnosis.POA-SVM was com-pared with GJO-SVM and PSO-SVM.The results show that POA-SVM has obvious improvement in fault rec-ognition rate and stability compared with GJO-SVM and PSO-SVM under different working conditions.
万方数据
维普
