Research On Remaining Life Prediction Of Railway Freight Car Wheels And Maintenance Strategy Optimization

In recent years,with the rapid development of China’s heavy haul railway,higher requirements are put forward for the maintenance technology of railway freight train.With the continuous development of Prediction and Health Management technology and the continuous improvement of on-line monitoring technology of railway freight train,it is an inevitable trend to apply the concept of condition based maintenance to the maintenance of railway freight train.As an important part of railway freight car,it is very necessary to accurately predict the remaining life of the wheel,and then determine the reasonable maintenance opportunity of the whole train wheel for ensuring the operation safety,improving the transportation efficiency and reducing the maintenance cost.In this paper,the wheel of the railway freight car is taken as the research object.The prediction of the remaining life of the wheel based on the on-line monitoring data and the optimization algorithm of the maintenance strategy are deeply studied.A opportunity centralized maintenance strategy of the wheel based on the prediction of the remaining life is proposed to optimize the wheel maintenance plan of the whole train.The main research work includes the following aspects:Firstly,the method of extracting wheel degradation trend components based on HP filter algorithm is studied.Due to the different positions of the wheels detected by TWDS system each time,the direct monitoring data of wheel wear fluctuates greatly.A method of extracting trend components of wheel degradation data based on HP filtering algorithm is proposed,and the trend components of degradation data are used in degradation process modeling and remaining life prediction.After HP filtering,the fluctuation of the data is reduced while the wheel wear degradation trend is retained.Then,a wheel tread wear degradation model based on nonlinear Wiener process is constructed and the parameters are estimated.According to the non-linear characteristics of the wheel degradation process,the wheel tread wear degradation model is constructed by using the non-linear Wiener process,and the model parameters are estimated by using Gibbs sampling algorithm according to the wheel historical degradation data.Next,the model parameter updating formula derivation and real-time remaininglife prediction are realized.In order to fuse the historical data with the information contained in the real-time monitoring data,the parameter updating formula of the drift coefficient in the nonlinear Wiener process model is derived by using the Bayesian formula.After the real-time model parameters are estimated,the real-time remaining life probability density function of the wheel is obtained according to the concept of the first arrival time of the stochastic process.Taking the expected value of the probability density function of the remaining life as the prediction value of the remaining life of the wheel,the real-time prediction of the remaining life of the wheel is realized.In order to prove the prediction effect of the model,the method proposed in this paper is compared with the existing method.The results show that the model and parameter updating method used in this paper have high prediction accuracy.After that,based on the prediction results of the remaining life of the wheel,this paper proposes an opportunity centralized maintenance strategy which combines the opportunity maintenance and the centralized maintenance,to solves the problem that the time of the centralized maintenance is difficult to be determined due to the difference of the degradation process of different individual wheels.The cost function reflecting the use cost of wheels is constructed,and the average cost per unit time and the number of temporary repairs are taken as the optimization objectives.The maintenance strategy optimization is realized by NSGA2 algorithm.In view of the poor diversity of the primary population,a semi random primary population generation method is proposed to improve the diversity of the primary population.Finally,based on a 500-day online monitoring data on a fixed group,consisting of54 freight train cars,the algorithm of remaining life prediction and maintenance strategy optimization proposed in this paper are verified.The results show that the algorithm proposed in this paper can achieve the optimal maintenance plan of wheels in a train according to the monitoring data of wheel degradation state.The optimized maintenance strategy can reduce maintenance times and maintenance costs.The method proposed in this paper is not only applicable to the formulation of wheel maintenance strategy for railway freight cars,but also to the formulation of maintenance strategy for other equipment with a large number of parts of the same type.