基于延迟补偿变分贝叶斯鲁棒滤波的重载列车编队状态估计
Delay-Compensated Variational Bayesian Filtering for State Estimation in Heavy-Haul Platooning with Unknown Noise
宋宗莹 1李烁 2王兴中 1杨迎泽3
作者信息
- 1. 国家能源集团,中国神华能源股份有限公司,北京 100011
- 2. 长沙理工大学 电气与信息工程学院,长沙 410114
- 3. 中南大学 计算机学院,长沙 410083
- 折叠
摘要
在重载列车编队运行中,对领航列车的位置与速度进行实时、精确的估计是实现列车自动防护（ATP）的关键。然而,车间通信延迟、丢包以及传感器噪声会严重削弱观测数据的可靠性。为保证恶劣工况下状态感知的可信度,本文设计并实现了一种变分贝叶斯鲁棒卡尔曼滤波器（VBRKF）,以联合估计系统状态与时变观测噪声协方差。具体而言,本文将观测噪声协方差建模为服从逆Wishart先验的随机矩阵,并将通信延迟等效为一种观测噪声,通过自适应调节滤波器对观测值与预测值的信任权重,实现对不断变化环境的动态适应。仿真结果表明,与扩展卡尔曼滤波（EKF）及传统VBRKF算法相比,本文的方法能够得到精度更高、更为稳定的状态估计结果。
Abstract
Accurate real-time state estimation of the leading train is essential to Automatic Train Protection (ATP) in heavy-haul platooning, where communication delays, packet loss, and sensor noise severely degrade observation reliability. This kind of state estimation includes position and speed estimation. To ensure trustworthy state perception under harsh conditions, this paper designs and implements a Variational Bayesian Robust Kalman Filter (VBRKF) that jointly estimates the system state and time-varying observation noise covariance. Specifically, we model the observation noise covariance as a random matrix with an inverse-Wishart prior, and then treat communication delay as a form of noise, finally adapting to changing environments by adjusting the trust in observations and predictions respectively. The simulation results indicate that this kind of VBRKF delivers more accurate, stable estimates compared to EKF and traditional VBRKF.关键词
重载列车编队/列车自动防护（ATP）/状态估计/变分贝叶斯滤波/通信延迟补偿Key words
Heavy-haul platooning/Automatic Train Protection (ATP)/State estimation/Variational Bayesian filtering/Communication delay compensation引用本文复制引用
宋宗莹,李烁,王兴中,杨迎泽.基于延迟补偿变分贝叶斯鲁棒滤波的重载列车编队状态估计[EB/OL].(2026-04-22)[2026-04-24].http://www.paper.edu.cn/releasepaper/content/202604-160.学科分类
铁路运输工程/自动化技术、自动化技术设备
评论