Distributed Optimal Control to Virtual Formation of Railway Trains With Dynamic Coupling/Decoupling: An Accelerated Projected Gradient Based Decomposition Method
Yiwen Zhang, Shukai Li, Lixing Yang
Abstract
To improve the operational efficiency of railway traffic systems, this paper designs distributed optimal control for railway trains based on the virtual formation technology. The objectives and safety constraints for train units under different operational conditions are proposed according to the communication structure of the virtual formation. Relative braking is adopted for train units inside the same formation, while absolute braking is used between different formations to achieve dynamic coupling and decoupling. Under the model predictive control framework, an optimal control model for each train unit with nonlinear coupling objectives and constraints is constructed. Successive linearization is employed to reduce the bias caused by model linearization. With the accelerated projected gradient based decomposition method, a distributed optimal control algorithm is designed to realize the real-time control for train units in multiple virtual formations. Numerical examples are provided to illustrate the effectiveness of the proposed model and method.