Residual life prediction of rolling bearing under multiple stresses

In order to realize the residual life prediction of rolling bearings under multiple stresses，a method for residual life prediction of rolling bearings based on acceleration model and Bayesian theory was proposed by effectively utilizing the degradation data under no stresses.The validity of the rolling bearing test data was analyzed by goodness of fit test and Weibull probability graph test.Switching Kalman filters （SKF） were used to determine the degradation state of the rolling bearing at each moment.When rolling bearing entered into the accelerated degradation，degradation process of bearing was fitted with exponential model，and the generalized linear logarithmic model was used to represent the relationship between degradation model parameters and stress；according to the revised bearing real-time degradation data，the model parameters were updated using Bayesian algorithm，the probability density function of the residual life of rolling bearing were acquired for residual life prediction of the rolling bearing.The XJTU-SY bearing data set was used for verification，and the root mean square error of the predicted results was less than 20 min，proving that the proposed method can effectively predict the residual life of rolling bearings.