A new binary wheel wear prediction model based on statistical method and the demonstration

Abstract

Wheel wear is an important factor that could affect the normal service of the high-speed trains. Wheel wear status plays a significant role in evaluating the safety and reliability in the operation of high-speed trains. At first, the line-tracking measurement was conducted and the wear profile data under different operation mileages was analyzed to get the characteristics of wheel wear. Then, polynomial fitting based on the least squares method was conducted twice to get a numerical wear prediction method in which the wear location on the profile and operation mileage was regarded as the variables. The accuracy of the prediction model was verified through the comparison of geometry characteristics and wheel/rail interaction. Utilizing the predicted and measured profiles, we found the consistent calculation results of vehicle ability, riding index and the characteristic curves of the lateral and vertical accelerations of key components. As a consequence, this binary model is an accurate approach for the application of wheel profiles prediction.