Research On Prediction Of Bearing Remaining Life Based On Condition Monitoring

Bearings are the key components of many rotating machinery and equipment.It plays an important role in the operation of mechanical equipment.Due to the complex operating conditions of bearings,it is one of the most prone to failure of mechanical equipment.The healthy operation of bearings is essential to ensure the safe and reliable operation of the entire equipment.Therefore,it is very necessary to monitor the condition of the bearing and predict the remaining life.In this paper,based on the measured bearing vibration signal as the basic data,the method of extracting and selecting the bearing degradation characteristics is studied.On this basis,the remaining bearing life prediction research is carried out.The main research contents of this paper are as follows:In view of the problem that the bearing degradation process is not easy to accurately characterize,this paper analyzes the characterization effect of the time domain,frequency domain and time-frequency domain energy eigenvalues on the bearing degradation process.The wavelet decomposition algorithm and the local mean decomposition algorithm are used to decompose the signal to obtain the time-frequency energy characteristic value of the bearing degradation process.The time-domain features and time-frequency energy features are extracted to form the bearing degradation feature set.Using the correlation coefficient method,the degraded features with better characterization performance are selected as the final eigenvalues.In view of the problem that the statistical model has poor predictive effect on individual bearings,this paper introduces a proportional hazards model combined with Weibull distribution to form a Weibull proportional hazards model with Weibull distribution failure rate function as the base function and bearing degradation characteristics as covariates to predict the remaining life of the bearing.The actual bearing vibration data is used to test the prediction effect of the model.The results show that the prediction result of the remaining life prediction method is relatively stable and meets the requirements of the application from a safety perspective.The problem of bearing degradation is not a single process but different degradation stages.In this paper,heuristic segmentation algorithm is introduced to divide the different stages of bearing degradation.The state change points that distinguish different degradation stages are obtained,and the information provided by the state change points is used to select the key stages in the bearing degradation process to construct a state space model of the bearing degradation process.The model parameters are updated using a particle filter algorithm that works well with non-linear systems to predict the remaining life of the bearings.