基于BP神经网络的捣固卫星小车制动距离预测方法研究

Research on Braking Distance Prediction Method of Satellite Car for Tamping Based on BP Neural Network

针对目前D09-32型捣固车自动捣固系统存在的卫星小车制动距离预测算法复杂以及工程实现难度较高等问题,文章提出一种基于BP神经网络的制动距离预测方法。该方法通过BP神经网络建立卫星小车制动速度、位置与对应的制动距离的关系模型,并利用所采集的卫星小车实时速度和位置数据预测卫星小车在当前速度与位置下的制动距离。对该预测方法关键模型进行Matlab仿真并与样本值进行对比分析,结果表明,采用所提出的基于BP神经网络的制动距离预测方法能够很好地对卫星小车制动距离进行预测,与现有的线性拟合预测方法相比,其预测误差平均减小了48.4%,有效提高了捣固车自动捣固的精度,且该方法具有良好的适用性和稳健性。

To deal with the obstacles in the automatic tamping system of D09-32 type tamping vehicles,such as the complicated braking distance prediction algorithm of satellite car and the challenges in engineering implementation,this paper proposes a braking distance prediction method based on BP neural network.By leveraging the data of real-time speed and position of the satellite car,the braking distance of the satellite car at the current speed and position is predicted by a BP neural network model representing the relationship between car's braking speed and position and its corresponding braking distance.The key model of this prediction method was simulated using Matlab software and the results were compared with the sample values.The comparison results show that the proposed braking distance prediction method based on BP neural network can satisfactorily predict the satellite car's braking distance.Compared with the currently prevailing linear fitting prediction method,the prediction error of this method is reduced by 48.4%on average,which can effectively improve the precision of automatic tamping,thus raising applicability and robustness of the tamping vehicles.