Research On Fault Diagnosis Of Rolling Bearing Based On Improved LSSVM

As the core component of rotating machinery and equipment,the performance of rolling bearing has a direct impact on the operation of machinery and equipment and system.Therefore,the work of fault diagnosis of rolling bearings can not be ignored.It can not only avoid unnecessary loss and maintenance cost to a large extent,but also improve the stability and safety of mechanical equipment.In this paper,according to the working principle of rolling bearings and the causes of its faults,according to the process of fault diagnosis analysis,starting from the fault vibration signals,the two key links in the process of rolling bearings fault diagnosis are studied and improved:fault feature extraction and fault type recognition.The main research work is as follows:In the process of fault feature extraction,aiming at the problems such as the difficulty in obtaining the fault features of rolling bearing vibration signals and the existence of modal aliasing in empirical mode decomposition,The Variational Mode Decomposition(VMD)based on Particle Swarm Optimization(PSO)is introduced,which can effectively solve the phenomenon of modal aliasing and avoid the objective effect caused by setting VMD parameters artificially.Firstly,the characteristics of vibration signals of rolling bearings are fully considered,and the envelope entropy which can characterize the characteristics of the vibration signals is selected as the fitness function of PSO,and then the optimal parameter combination(K,α)of VMD is obtained.The feasibility and effectiveness of VMD optimization based on PSO algorithm are demonstrated by simulation and measured signal analysis experiments.Finally,K Intrinsic Mode Function(IMF)components obtained by VMD decomposition were analyzed and calculated by combining VMD with multi-scale Permutation Entropy(MPE),so that the fault features could be extracted comprehensively and the fault types could be identified as the feature vectors of the fault signals.In the process of fault type identification,aiming at the difficult to obtain a large number of diagnostic data and the method of selecting model parameters is easy to fall into local optimum,Least Squares Support Vector Machine(LSSVM)was used to recognize the fault types.Considering that the selection of regularization parameter c and kernel width σ of LSSVM will affect its classification performance,Bat Algorithm(BA)is used to optimize the parameter combination of LSSVM,the convergence experiments show that the BA optimization algorithm converges faster and shows good convergence precision and stability compared with PSO and GA.Finally,a rolling bearing fault diagnosis model based on BA-LSSVM was proposed.The recognition of actual fault types was used to verify the accuracy and reliability of the BA-LSSVM model,and a comparison model was built.The experimental results show that the BA-LSSVM model can effectively realize the classification of rolling bearing fault types,and the accuracy of the classification is high.