Research On Fault Diagnosis And Remaining Life Prediction Method Of Rolling Bearings

Rolling bearings are the core component of mechanical equipment,their working status directly determine the overall performance of the equipment and they have an irreplaceable position in the application of mechanical industrial products.Fault diagnosis and remaining life prediction of rolling bearings can avoid accidents of mechanical equipment and provide safety and reliability decision support for equipment maintenance plans.How to effectively diagnose the fault of rolling bearings and predict the remaining life of rolling bearings is of great significance for the health management of mechanical equipment.Based on the original vibration signals collected by the test platform,this paper studies the fault diagnosis and remaining life prediction method of rolling bearing based on the data-driven method.The main contents are as follows:(1)Aiming at the problem that the fault features of rolling bearing are difficult to extract,a fault diagnosis method for rolling bearing based on parameter optimized variational mode decomposition(VMD)is proposed.The number of modal component and penalty factor of VMD decomposition algorithm are difficult to determine,this paper proposes a method to optimize the two parameters by using differential evolution(DE)algorithm and grid search algorithm.The simulated signals and measured signals of the rolling bearing are analyzed according to the parameter optimized VMD decomposition algorithm,which prove the feasibility and effectiveness of the proposed method.(2)The feature extraction characterizing the rolling bearing performance degradation states and the establishment of remaining life prediction model are the premises for developing the remaining life prediction of rolling bearing.Taking the bearing vibration data of the accelerated test platform as the research object,firstly,the original vibration signals are denoised and the time domain features,frequency domain and wavelet packet energy features are extracted.Then use the correlation coefficient and mutual information to reduce the features to form the feature set of the prediction model.Finally,BP neural network(BPNN)remaining life prediction model is established,use the particle swarm optimization algorithm to optimize weights and thresholds of BP neural network,predict the remaining life of the test bearing by using the optimized model,and compare the effects of two feature reduction methods on the prediction of remaining life,the results prove that the features selected by correlation coefficient are more effective.(3)In order to accurately predict the remaining useful life of rolling bearings,the fuzzy comprehensive evaluation method is used to solve the problem that the bearing degradation state is difficult to divide,with the help of the support vector machine which has good ability in small sample data analysis and efficient parallel search for information of differential evolution algorithm,a remaining life prediction model for rolling bearings based on support vector regression and differential evolution algorithm is proposed.The model uses the differential evolution algorithm to get the optimum parameters of the support vector regression model,the feasibility of the remaining life prediction model is verified by the accelerated life testing data collected by the test platform.