Accurate Modeling of Railway Turnout Systems and High-Precision State Estimation Methods

The railway turnout system (RTS) is a critical component of modern, complex railway transportation networks, where its reliability and safety are essential for ensuring operational efficiency and transportation safety. To address the issue of inaccurate modeling in current turnout systems, this article establishes a precise RTS model based on bond graph (BG) theory that incorporates the sensor model, while considering the effects of gear backlash and nonlinear friction. A nonlinear state-space model of the RTS is further developed by integrating it with a constructed measurement model. Additionally, this article introduces an improved extended Kalman filter (EKF) design method for high-precision state estimation of the nonlinear model. This filter incorporates second- and third-order statistical information derived from nonlinear partial Taylor expansions, significantly enhancing estimation accuracy. In the experimental section, the impact of nonlinear factors on model accuracy is analyzed by comparing data from the model’s detectors with actual sensor data, utilizing ablation experiments. Furthermore, the superior performance of the proposed method in nonlinear filtering is validated through the estimation results of primary state variables in the turnout system.