基于贝叶斯建模的轨道占用识别方法

识别列车所在轨道，是列车运行控制系统必不可少的功能. 提出一种数字轨道地图辅助的基于贝叶斯建模的轨道占用识别方法. 首先，在考虑全球导航卫星系统(Global Navigation Satellite System, GNSS)与方向相关测量误差的基础上进行地图匹配，采用卡尔曼滤波算法处理速度测量值，通过加权求和的方式对GNSS与速度信息进行融合，获得列车沿轨道方向的一维位置；其次，对列车位置假设进行贝叶斯建模，计算在给定GNSS与速度测量的前提下所有可能的位置假设的概率；最后，与设置的假设概率阈值进行比较，对不同的识别结果进行分类. 实验结果表明，基于贝叶斯建模的轨道占用识别法能够减少剔除小概率假设所需的距离，与垂直投影法相比，该方法可以对列车所在轨道做出更确定的判断.

A Track Occupancy Identification Approach Based on Bayesian Modeling

The identification of the present track is an essential applications of train control systems. This paper presents a track occupancy identification approach based on Bayesian modeling with the aid of a digital track map. Firstly, map matching is performed considering the direction-dependent GNSS standard deviation while velocity measurements are processed using Kalman filtering. The longitudinal train position is obtained by fusing GNSS and velocity information by means of weighted averaging. Then a Bayesian model is built for hypotheses of the train position and the probabilities of all possible hypotheses given specific GNSS and velocity measurements are calculated. Finally, the probabilities are compared with a defined threshold. Moreover, the identification results are classified. The experimental results show that the track occupancy identification approach based on Bayesian modeling can reduce the distance to eliminate hypotheses with negligible probabilities. Compared to an orthogonal projection algorithm, this approach is more confident in the determination of the track that the train takes.