摘要
以轨检车测出的左轨和右轨轨向不平顺、左轨和右轨高低不平顺为输入,以轨检车测出的轮重减载率为输出,采用贝叶斯正则化算法构建了NARX神经网络。仿真试验结果及与BP神经网络输出结果的对比表明,采用NARX神经网络实现轮重减载率预测是可行的,NARX神经网络比BP神经网络更适用于减载率预测。
The track alignment irregularity of the left rail and right rail, the track vertical profile irregularity of the left rail and right rail measured by the rail inspection car are used as input, and the reduction rate of wheel load measured by the rail inspection car is used as the output, the Bayes regular calculation method is applied to construct the NARX network. The comparison between the simulation test result and the BP neural network output result shows that it is feasible to use the NARX network to realize the prediction of reduction rate of wheel load, and the NARX network is more appropriate to the prediction of reduction rate of wheel load than the BP neural network.
出处
《铁道车辆》
北大核心
2012年第9期4-7,1,共4页
Rolling Stock
基金
科技支撑项目资助(2011BAG01B05)