Nonlinear model predictive control to automatic train regulation of metro system: An exact solution for embedded applications

Abstract

This paper investigates the nonlinear model predictive control to automatic train regulation in metro systems under dynamically disturbed environments. A coupled state-space model for the evolution of train traffic and passenger load dynamics is explicitly developed. With the safety constraints, a mixed-integer nonlinear optimal control model is formulated to generate train regulation strategies. An efficient solution algorithm, based on spatial branch and bound combined with generalized Benders decomposition, is particularly designed to obtain exact solutions quickly for the embedded applications. The original complex problem can be converted into several smaller convex quadratic subproblems with reduced domains to be solved quickly. The effectiveness of our train control model and approaches are verified in numerical simulations. Computational results illustrate that the proposed control method effectively improves the operational efficiency and meanwhile reduces energy consumption during operations. Besides, the efficient embedded solution algorithm performs satisfactory computational efficiency, facilitating a real-time implementation.