轮轨接触点的在线连续测量  

On-Rail Continuous Measurement of Wheel/Rail Contact Point

作　　者：陈建政[1] 王志强 

机构地区：[1]西南交通大学牵引动力国家重点实验室,四川成都610031 [2]宁夏地球物理地球化学勘查院,宁夏银川750004

出　　处：《中国铁道科学》2007年第5期15-18,共4页China Railway Science

基　　金：国家自然科学基金资助项目(50375129);牵引动力国家重点实验室开放基金资助(TPL0607)

摘　　要：车辆运行过程中,由于车轮旋转的影响,轮轨作用点位置的测量非常困难。在线测量作用点的轮轨位置对脱轨机理和机车车辆性能的研究有十分重要的意义。在常规测力轮对的基础上,通过增加1个电桥感应作用点位置的变化。对电桥的输出进行傅立叶级数分析,建立作用点位置与电桥输出的非线性方程组。针对非线性方程组难以实时求解的问题,采用神经网络拟合轮轨作用力位置变化与电桥输出间复杂的非线性映射关系,用不同作用点位置下各种横、垂向力的组合对神经网络进行训练,以达到求解非线性方程组、得到作用点位置的目的。实验结果表明,可以较为准确地检测轮轨力作用点的位置,而且预测能力也令人满意。It＇s very difficult to detect wheel/rail contact points of the running vehicle because of wheel rotation during railway vehicle operation. On-rail measurement of wheel/rail contact points is of great importance to the research of derailment mechanism and vehicle performance. On the basis of conventional instrumented wheelset, an electric bridge to induce the changes of contact point is added. Nonlinear equations between the contact point and the outputs of the bridges are set up by means of the theory of Fourier series. As it is difficult to work out real-time solutions for nonlinear equations, a method based on artificial neural network is studied to solve these equations. Neural network is trained to learn the non-linearity relationship between wheel/rail contact point and the bridge outputs, with the combination of various horizontal and vertical loads at different positions. Experiment results show that the position of contact point can be obtained more accurately and the forecast ability of the neural network is satisfactory.

关 键 词：轮轨作用点 傅立叶级数 神经网络 测力轮对 

分 类 号：U270.1[机械工程—车辆工程]