Research On Preventive Maintenance Of Rolling Bearing Based On Improved Animal Migration Algorithm

In the process of maintenance decision,the fault diagnostic of bearing is one of the most common process.Bearing test is mainly used in rolling bearing fault detection.Its concrete working status directly affects the machine working smoothly and play a decisive influence on maintenance decision.So the research of rolling bearing fault diagnostic technology is of great practical significance and practical value.This paper is mainly about the fault feature extraction algorithm and fault diagnostic methods.The main contents of the paper are as follows:Aimed at the weakness existed in the animal migration method in parameter estimation.The method of artificial fish algorithm of parameter estimation was used to the animal migration method,so its optimization ability and convergence speed were improved.Then the more stable model parameters were gotten and the precision of model identification was improved.Using the empirical mode decomposition(Empirical Mode Decomposition,EMD)method,multiple sets of rolling bearing signal were divided into different frequency distribution of intrinsic function(Intrinsic Mode Function,IMF).Then,rich fault signal of the top five IMF were concluded and calculated,and kurtosis value and composition of fault feature vector were gotten.Then the fault characteristic vectors were viewed as input and put into improving animal migration identification method,and working state and fault type were classified.On the basis of the envelope sample entropy,the kurtosis value of IMF was calculated,the IMF kurtosis value decomposition was applied to the analytic target vibration signal by using EMD method.Then the richly fault characteristics of the IMF was solved into the corresponding envelope signal.Finally,the signal was separated and calculated effectively by using sample entropy method.The envelope signal was optimized to calculate the sample entropy feature vector.The fault classification was forecasted by using the SVM classifier.The use of machine work data,improved animal migration algorithms,and envelope-based feature signals IMF kurtosis values were used to calculate the SVM(support vector machine)fault signal classification.The results of the experimental signal show that the improved animal migration algorithm helps the fault signal analysis.The envelope signal can be used to calculate the entropy of the envelope,and the fault can be diagnosed effectively by SVM.