基于BP神经网络的钢轨磨损量预测

Prediction of Wear Volumes of Rail Steel Based on BP Neural Network

随着列车运行速度和轴重的提高,轮轨系统的磨损越来越严重,其中曲线半径、轴重和运行速度是影响轮轨磨损的重要因素。建立了钢轨磨损量影响规律的径向BP基函数神经网络模型,该网络具有3路输入,3个神经层;在JD-1大型轮轨模拟试验机上通过改变试验参数进行钢轨磨损试验,获得不同试验参数下的钢轨磨损量;以钢轨磨损数据作为BP神经网络的目标样本,对不同试验参数下的磨损量进行了预测。结果表明,模型可较准确地计算轮轨冲角和速度对钢轨磨损量的影响规律,利用BP神经网络对钢轨磨损量预测具有较高的精度,可在一定程度上验证试验结果。

With increasing the speed and axle load of railway, the wear of wheel/rail is becoming more and more serious. The curve radius, axle load and running speed are main influence factors of the wear of wheel/rail. The BP neural network used in the forecast of the rail wear volume was established. The network has three input vectors and three neural layers. The rail wear volume was obtained using JD-1 rail-wheel tribology simulating machine under different experimental parameters. The wear volumes of different experimental parameters were forecasted using BP neural network. The results indicate that it is feasible to forecast the effect of the angle of attack and speed between wheel and rail on rail wear volume. The predicted results have high precision and it can verify experimental results in a certain extent.