Railway system performances depend on effective dynamic scheduling and train operation control. The fast expansion and increasing complexity of high-speed railway (HSR) networks raise new challenges in maintaining the punctuality and efficiency in daily operations, in particular, in the event of disruption. This paper aims to review the state-of-art in dynamic traffic scheduling, trains operation control, and their integration for safer, more punctuate, efficient, and resilient HSRs, whose origins may trace back to their counterparts in traditional railways. First, the existing two-tier hierarchy of scheduling and control in HSR’s daily operation is introduced. At the higher layer of scheduling, a general model of dynamic train scheduling is discussed, followed by reviewing the scheduling methodologies. At the lower layer of train operation control, recent progress in tracking control of high-speed trains is discussed, with focus on the latest advances in single train control and cooperative control for multiple trains. Then, as the trend of technological progress for future HSRs, the recent development of integrating dynamic scheduling and operation control is introduced, which is made possible by efficient information exchanges among the scheduling subsystem and the train control subsystem. A three-layer integration framework and associated co-optimization methodologies are presented by introducing a co-optimization layer that bridges the separated scheduling and control layers. Finally, this review is concluded with discussions on open questions and possible directions for future research.

Train Timetable Rescheduling (TTR) is a crucial task in the daily operation of high-speed railways to maintain punctuality and efficiency in the presence of unexpected disturbances. However, it is challenging to promptly create a rescheduled timetable in real time. In this study, we propose a reinforcement-learning-based method for real-time rescheduling of high-speed trains. The key innovation of the proposed method is to learn a well-generalized dispatching policy from a large amount of samples, which can be applied to the TTR task directly. At first, the problem is transformed into a multi-stage decision process, and the decision agent is designed to predict dispatching rules. To enhance the training efficiency, we generate a small yet good-quality action set to reduce invalid explorations. Besides, we propose an action sampling strategy for action selection, which implements forward planning with consideration of evaluation uncertainty, thus improving search efficiency. Extensive experimental results demonstrate the effectiveness and competitiveness of the proposed method. It has been proven that the local policies trained by the proposed method can be applied to numerous problem instances directly, rendering it unnecessary to use human-designed rules.

Train timetable rescheduling (TTR) aims to promptly restore the original operation of trains after unexpected disturbances or disruptions. Currently, this work is still done manually by train dispatchers, which is challenging to maintain performance under various problem instances. To mitigate this issue, this study proposes a reinforcement learning-based approach to TTR, which makes the following contributions compared to existing work. First, we design a simple directed graph to represent the TTR problem, enabling the automatic extraction of informative states through graph neural networks. Second, we reformulate the construction process of TTR’s solution, not only decoupling the decision model from the problem size but also ensuring the generated scheme’s feasibility. Third, we design a learning curriculum for our model to handle the scenarios with different levels of delay. Finally, a simple local search method is proposed to assist the learned decision model, which can significantly improve solution quality with little additional computation cost, further enhancing the practical value of our method. Extensive experimental results demonstrate the effectiveness of our method. The learned decision model can achieve better performance for various problems with varying degrees of train delay and different scales when compared to handcrafted rules and state-of-the-art solvers.

This article presents a novel disturbance decoupling generalized proportional–integral observer (DD-GPIO) for the early detection of sensor faults in dynamic systems subject to external semistationary periodic disturbances. The focus is to reduce the adverse impacts caused by the disturbance whose spectrum is uneven with some dominant frequencies. This achieved in a two-stage design procedure. In the first stage of zero-assignment, the additional zeros introduced by the state transition matrix of GPIO are allocated near the disturbance frequencies to construct a rank-deficient disturbance transfer function matrix (DFTM) relating from the disturbances to the fault detection residuals. In the second stage, an additional gain matrix is applied to the output estimation error before the error is fed into the GPIO. The additional matrix can be designed by using the eigenvector of zero eigenvalue of the rank deficient DFTM, such that the periodic perturbations can be completely decoupled from the fault detection residual for better performance in fault detection. The conditions to achieve disturbance decoupling is given. Finally, a novel condition number-based sensitivity objective is proposed to optimize the DD-GPIO's poles. Simulation and a two-wheeled robot experiment illustrate the outperformance of the DD-GPIO in decoupling periodic disturbances and detecting minor sensor faults.

This article investigates the detection and identification of physical faults in devices and false-data-injection attacks in communication networks for multi-agent systems with event-triggered transmission mechanisms and subject to external periodic disturbances. First, a new detection and identification scheme, including a local disturbance decoupling (LDD) observer and a distributed disturbance decoupling (DDD) observer, is proposed. Then, based on zero-assignment and the rank-deficiency of the transfer function matrix at zeros, a co-design method for the LDD observer and DDD observer is proposed, which enables the decoupling of periodic disturbances from the residuals for detection and identification. This new scheme no longer requires the transmission of control signals from the node being monitored or the exchange of information between its neighbors, significantly reducing the communication overhead and enhancing the system's security. Finally, a simulation based on a multi-two-wheeled trolley system is used to verify the effectiveness of the proposed method.