Research on Optimization of Wheel Repair Strategy Method Based on ARIMA-random Forest for EMU
Based on the characteristics of EMU running mileage and the degradation of wheel as non-stationary time series,the CR400BF was used as the research object.The ARIMA and random forest algorithm are combined to carry out the wheel degradation prediction and lathing strategy optimization research.ARIMA is used to carry out differential processing on the running mileage data,construct random forest decision tree based on Gini coefficient partition characteristics,input wheel detection history data into training set and test set for training,determine the predicted value of wheels size with the predicted mean value.Constrained by the geometric size and dynamic performance of the wheel set,the longest service life of wheel and the minimum number of lathing and ride index as the optimization objectives,establish the optimization model of wheels lathing strategy,and the lathing amount of wheels and the wheel diameter value are predicted.The results show that when the wheel repair diameter amount is 2.5 mm,the edge of wheel thickness is HAi=28.5 mm and HBi=30 mm,the lathing strategy is the best,and the wheels life can be increased by 31.4%.The research results can provide theoretical support for the optimization of wheelset rotation repair strategy of EMU.
万方数据
维普
