| During the operation of mechanical equipment,if the rolling bearing fails,it will bring serious consequences.Therefore,it is very important to strengthen the research on the fault diagnosis method of rolling bearing.This paper mainly studies the fault diagnosis method of BP neural network.Aiming at the problem of low accuracy,according to the fault diagnosis process,this paper optimizes the three stages of vibration signal analysis,fault feature extraction and fault diagnosis,so as to improve the accuracy of fault diagnosis.In the first stage,the vibration signal analysis process is optimized and the vibration signal processing is strengthened.In this paper,correlation masking empirical mode decomposition(CMEMD)is proposed.This method can not only solve the problem of modal aliasing,but also complete the first step of screening IMF components and select more valuable IMF components.Through experiments,the time-frequency analysis of vibration signals in this stage and the verification of CMEMD method are completed,which fully proves the effectiveness of CMEMD method.In the second stage,the fault feature extraction process is optimized to determine a more effective and representative fault feature vector.This paper proposes a double entropy method(DE),which combines the sample entropy and Shannon entropy to jointly participate in the process of determining the fault feature vector and complete the second step of screening the IMF component.The correlation masking empirical mode decomposition method is combined with the double entropy method to determine the fault feature vector.Through experiments,the verification of double entropy method and the determination of fault feature vector are completed.The experimental results show the effectiveness of the double entropy method and the fault feature vector determination method.In the third stage,the fault diagnosis process and BP neural network fault diagnosis method are optimized,and combined with the optimization methods in the first two stages to improve the accuracy of fault diagnosis.In this paper,dual strategy particle swarm optimization(DSPSO)is proposed to optimize BP neural network,and a dual strategy particle swarm BP neural network bearing fault diagnosis method is proposed.Through experiments,the verification of the dual strategy particle swarm optimization algorithm in this stage is completed,and the diagnosis results are compared with the diagnosis results of BP neural network fault diagnosis method,which proves the effectiveness of each stage optimization method and the new bearing fault diagnosis method. |
