高速铁路列控系统列车位置在线估计算法研究

摘要

To address the train positioning error of the train control system of high-speed railway,an analysis was first made on the data of the distance error correction of different types of ATP onboard equipment through the absolute positioning information of the balise,to verify the characteristics of the normal distribution of the relative error of train positioning.Then an approximate model was proposed for range acquisition with the de-tecting data from onboard equipment,which was online identified by least square method.A discrete linear model for updating relative position of train was built and the stack denoise autoencoder aided Kalman filter was used to carry out online estimation of the relative position of the train to obtain the absolute position integrating the location of the balise.Based on the field data from type test of ATP onboard equipment and high-speed train,emulation experiments were conducted.Emulation results showed that the method proposed in this paper can reduce the positioning error effectively during the operation of high-speed trains.Precision improved at least 34.39％,based on corresponding root mean square error statistics.%针对高速铁路列控系统的列车定位误差问题,首先分析不同型号ATP车载设备通过应答器的绝对定位信息进行距离误差校正的数据,验证列车定位相对误差的正态分布特性;其次提出列车的相对测距近似模型,并以最小二乘法进行在线参数辨识;最后建立列车位置更新的离散线性模型,以定位相对误差的噪声特性和最小二乘的相对测距近似模型为基础,利用多层降噪自动编码网络辅助的卡尔曼滤波算法直接对列车的相对位置在线估计从而得到绝对定位信息.利用ATP车载设备与高速动车组接口型式试验的现场数据进行仿真分析,仿真实验与实际误差校正的对比结果表明,文章提出的方法可以有效降低高速列车行走过程中的定位误差,对应均方根误差统计,整体改善程度至少达到34.39％.