Research And Implementation Of Mileage Life Prediction Of Heavy-Duty Railway Freight Car Wheels

Since the "Thirteenth Five-Year Plan",with the steady growth of China's social economy,heavy-duty railway trucks have developed rapidly.However,the increasing speed and carrying capacity of the railway has brought tremendous pressure on the railway operation and maintenance personnel.At the same time,in view of the old-fashioned operation and maintenance model,it is necessary to make a "great upgrade" to the existing railway operation and maintenance.Plan to repair and change to "state repair."The truck wheel is one of the most important parts of the truck.The damage of the wheel seriously affects the safety of the truck.Therefore,the heavy-duty railway wagon wheel is selected as the research object,and the relationship between the wear mechanism of the wheel and the wheel mileage is studied.A Lean Square SVM(LSSVM)wheel mileage life prediction model based on the grey wolf optimization algorithm is used to accurately predict the mileage life of heavy-duty railway wagon wheels.The work of this paper is briefly described as follows:(1)Firstly,the existing indicators that can measure the wheel life are analyzed.It is found that these indicators can not meet the "state repair" mode of heavy-duty railway trucks.This paper studies and selects a new indicator to measure the wheel life,namely the wheel mileage.Lifetime,and by analyzing the characteristics of wheel wear data and existing engineering conditions,Support Vector Machines(SVM)was selected as the model for wheel mileage life prediction.(2)Then combing the wheel wear data set of the existing railway system,and found that it can not meet the requirements of the wheel mileage life prediction system.Based on the characteristics of the wheel wear data set of heavy-duty railway wagons,this paper starts with the complicated data of each railway system.The parts history data and the component design life data are separately selected from the corresponding data systems,and combined with data cleaning and fusion methods,a suitable data set that can be directly applied to the truck wheel mileage life prediction system is constructed;Based on the constructed data set,LSSVM in SVM theory is selected as the mileage prediction model for truck wheels.And it is compared with the SVM life prediction model.The experimental results show that the SVM will greatly increase the training time of the training model under the condition of too much sample data,which leads to the low efficiency of the algorithm training,and the LSSVM has obvious advantages in training efficiency.(3)However,the parameter optimization ability of LSSVM is still not strong enough.For the defect of LSSVM optimal parameter finding ability,the LSSVM is improved by the grey wolf optimization algorithm to improve its parameter optimization ability and obtain a more accurate prediction model.Name it GWO-LSSVM prediction model.Based on the constructed prediction model,the actual wheel mileage data is verified.The results show that the GWO-LSSVM algorithm is advanced in parameter optimization.(4)Finally,the wheel mileage life prediction model is integrated into the "state repair"diagnosis decision comprehensive judgment system.