Time-varying Parameter Degradation Model For Remaining Useful Life Prediction Of Rolling Bearing

As one of the key components of industrial equipment,rolling bearings are affected by uncertain factors such as environmental conditions and operating conditions in practical engineering applications,and various failure modes may occur during the operation of the bearing,resulting in non-stationary vibration signals.For the prediction of the remaining life of rolling bearings under different working conditions,on the one hand,it is difficult to accurately extract the health index of the non-stationary signal caused by the variable speed.On the other hand,the parameters of the bearing degradation model change with the degradation of the bearing performance.The remaining life prediction brings a large error.In response to the above problems,this thesis firstly calculates the fault characteristic frequency by analyzing the failure mechanism of the bearing.Secondly,use different feature extraction methods to obtain different feature information,and then assess the various degradation features of the bearing through the evaluation index,analyze the influence of different features on the bearing degradation,and thus accurately extract the sensitive features representing the operating state of the bearing throughout its life cycle,the optimal degradation feature index can be obtained.For constant working conditions,a method for predicting the remaining life of rolling bearings based on a time-varying parameter model and particle filtering is proposed.Firstly,the SVDD model is used for training and testing to determine the starting point of life prediction;secondly,the time-varying parameter degradation model of the bearing is established according to the analysis results of the real-time finite element model,and the particle filter algorithm(PF)is used to update and estimate the degradation state of the bearing.Predict its remaining life.Finally,the proposed method is verified by using the XJTU-SY bearing public data set and the experimental data of the BPS rolling bearing accelerated life prediction test bench,and compared with the prediction results based on the traditional empirical model method.The results show that the method proposed in this paper has the highest residual life prediction accuracy compared with other models.For the variable speed condition,a residual life prediction model of angular-domain unscented particle filter rolling bearing based on time-varying degradation model parameters is proposed.First,the rolling bearing signal under variable rotational speed is transformed from the time domain to the angular domain by using order analysis technology,and then the health index of the bearing is extracted from the angular domain signal;The parameters of the performance degradation model are updated according to the calculation results of the finite element method;then,combined with the bearing performance degradation model with timevarying parameters,the unscented particle filter algorithm is used to track the bearing degradation state and predict its remaining life.Finally,the proposed method is verified by using the XJTU-SY bearing public data set and the experimental data of the BPS rolling bearing accelerated life prediction test bench,and compared with the prediction results of the traditional empirical model and the AR model method.The results show that under the condition of variable speed,the method proposed in this paper can more accurately predict the bearing life and obtain better performance than other traditional degradation model methods.