Real-time optimization of train regulation and passenger flow control for urban rail transit network under frequent disturbances

Abstract

Urban rail transit networks face frequent delays and oversaturation resulting from unexpected disturbances and high demands during rush hours, which inevitably leads to unstable train operations, poor service experiences, and potential accidents on crowded platforms. This paper explores the train regulation and passenger flow control strategies for large-scale urban rail transit networks. Specifically, we develop a mixed-integer nonlinear programming (MINLP) formulation to improve the punctuality and regularity in train operations, reduce the passenger waiting time, and alleviate the passenger flow burden of platforms, where the dynamic coupling interactions among network-wide passengers, trains and stations are systematically considered. To reduce the computational complexity and satisfy real-time requirements, we particularly devise a real-time optimization approach based on iterative nonlinear programming (INP) combined with Lagrangian relaxation (LR) under the rolling horizon (RH) framework, which effectively disposes of the intractable nonconvexity and constructs the desirable properties of decomposability and parallelism. The original problem is transformed into a sequence of line-level subproblems that can be solved quickly, so as to circumvent the computational burden. Moreover, to verify the effectiveness of our approach, we apply it to address numerical examples on a simplified network and the real-world Beijing Subway network. The realistic case studies illustrate that the proposed approach can reduce the train deviation from the original timetable, the passenger waiting time, and the passenger accumulation risk on the platform, by around 84.92%, 30.34% and 61.42% in comparison with the common rule-based strategy. Additionally, the computation time for the realistic large-scale experiment is acceptable for a real-time implementation.