关键词: 铁路运输, Koopman算子, 模型预测控制, 高速列车, 扩张状态观测器, 速度控制

Abstract: Targeting the topic of high-speed train speed control amid external disturbances, a modeling method of a high-dimensional linear model of high-speed trains based on the Koopman operator is proposed, and a composite controller (ESO-K-MPC) combining extended state observer (ESO) and model predictive control (K-MPC) based on Koopman operator is designed. Firstly, the dynamic high-dimensional linear model of a high-speed train with dynamic nonlinear characteristics was developed by approximating the infinite-dimensional linear Koopman operator using the extended dynamic mode decomposition algorithm. Secondly, the model predictive control was implemented, and the extended state observer was built to estimate and correct the system's total disturbance. The ESO-K-MPC-based highspeed train speed control system was created, together with the controller and control algorithm. Finally, combined CRH3 train parameters with the actual line data of Huashan North Station, Xi'an North Station of Zhengzhou-Xi'an High-speed Railway, the designed control methods and algorithms were simulated and studied without disturbance and white noise interference respectively. The simulation indicates that the displacement and velocity forecast accuracy of high-speed train modeling based on Koopman is increased by 83.86 and 87.40%, respectively, when compared to the linear state space model. ESO-K-MPC can accurately estimate and compensate interference during high-speed train operation, the control output curve practically overlaps the expected curve, and high precision tracking of the predicted curve during train operation is achieved.

Key words: railway transportation, Koopman operator, model predictive control, high-speed train, extended state observer, speed control