基于抗差自适应滤波的高速列车融合测速算法

Fusion Speed Measurement Algorithm of High-speed Train Based on Robust Adaptive Filter

摘要:

针对Kalman滤波在高速列车融合测速过程中因观测粗差和动力学模型误差而引起的融合精度下降问题，提出一种基于抗差自适应滤波的高速列车融合测速算法。首先，在Kalman滤波的基础上构建异常检测函数和误差判别统计量，用于检测和区分传感器异常观测导致的观测粗差和动力学模型误差；然后，针对观测粗差和动力学模型误差，分别采用三段式函数和指数函数构造抗差因子和自适应因子，通过2种因子合理调节观测信息和模型信息在状态估计中的权重，从而降低观测粗差和动力学模型误差对融合结果的影响；最后，通过2种运行场景以及算法对比，仿真验证抗差自适应滤波算法性能。仿真结果表明，与基于Kalman滤波的融合测速算法相比，所提出算法无论在观测粗差场景还是在动力学模型误差场景，均具有更高的精度和稳定性。

Abstract:

A fusion speed measurement algorithm of high-speed trains based on robust adaptive filter was proposed to solve the problem that the fusion accuracy decreased due to the observation gross errors and the dynamic model errors in the fusion speed measurement using Kalman filter. Firstly， the anomaly detection function and error discrimination statistics were constructed on the basis of Kalman filter， which were used to detect and distinguish the observation gross errors and dynamic model errors caused by abnormal observations of sensors. Then， for observation gross errors and dynamic model errors， a three-segment function and an exponential function were used to construct robust factor and adaptive factor， respectively. The weights of observation information and model information in state estimation were reasonably adjusted by the two factors， so as to reduce the impact of observation gross errors and dynamic model errors on the fusion results. Finally， the performance of robust adaptive filter was verified by simulation with two operation scenes and comparison between algorithms. The simulation results show that compared with the fusion speed measurement algorithm based on Kalman filter， the proposed algorithm has higher accuracy and stability in both the observation gross errors scene and the dynamic model errors scene.