Research On Predictive Maintenance Strategy Of EMU Brake Pads Based On Multivariate Integrated Random Walk

With the large-scale comprehensive development of the high-speed railway globally,the security and comfortableness of the Electric Multiple Unit(EMU)operation have been attached to more importance.And the increasing operating speed is a major test for the braking performance of the EMU.The brake pad is one of the important components of the basic actuator in the brake device,its price is high and the number of installations on the train is numerous,which directly determines whether the EMU can operate safely and reliably or not.The changeable environment and the harsh climatic condition along the railway line have a serious impact on the performance degradation of the brake pad.Safe and economical predictive maintenance strategy can guide the manager to take timely and reasonable maintenance for the brake pad,which can ensure the safe operation of the EMU and reduce the total cost of operation and maintenance.Based on the Multivariate integrated random walk model,the state space model of the trend of the brake pad's performance degradation is established in order to solve the problem that the low accuracy of the state estimation of the brake pad which is resulted from various random factors in the changeable operating environment.The Kalman filter theory is used to estimate and predict the future degradation of the brake pad.In order to determine whether the abnormal condition occurs or not before the critical degradation failure threshold is reached,the tolerance range of abnormal failure is determined.And then the reliability of the brake pad is predicted and the expected maintenance cost model is established.On this basis,the predictive maintenance strategy for the brake pad is developed.The theoretical analysis and simulation results show that this method is correct and effective.The main contents of this thesis are as follows.(1)Combined with the actual wear situation of the brake pad,the various factors that affect the performance degradation of the brake pad and the characteristics of the MIRW model are analyzed.Considering the decreased trend of the data in the next time due to the continuous performance degradation of the thickness observation series in the actual operation,the state space model of the brake pad's performance degradation is established based on MIRW.In this method,brake pad's performance degradation trend is extracted by fixed interval forward-backward smoothing algorithm firstly,and then on this basis,the future degradation state is predicted and estimated by the Kalman predictor.(2)Based on the obtained state estimation value,the tolerance range of abnormal failure is determined by the Grubbs criterion and distance discriminant theory to judge whether the brake pad is in abnormal condition before reaching the final failure.According to the reliability theory and maintenance theory and other aspects of the relevant knowledge,the cumulative failure probability and reliability is obtained.And then the expected maintenance cost model was established considering the accumulated cost of losses arising from the continuous degradation.According to the excepted total profit loss cost either maintenance or not,the predictive maintenance strategy for the brake pad is developed.(3)Taking the brake pad thickness data of a certain type of the CRH EMU as a practical case,the simulation is carried out.The analysis and comparison show that the method in this thesis can improve the accuracy of the thickness estimation and determine timely whether the abnormal situation occurs or not.It demonstrates that it meets the security and cost-effective requirements in China's EMU maintenance system status currently.