The shunting scheduling of EMU first-level maintenance in a stub-end depot

While solving the shunting scheduling of EMU first-level maintenance (SSEFM), most existing literature assumed a single maintenance route for all trains and considered only a through depot. It neglects the problem-specific characteristics in terms of varied maintenance routes and a stub-end depot, causing the infeasibility of the generated schedule in such particular circumstances. Therefore, the SSEFM problem with flexible maintenance routes in a stub-end depot with a transversal yard configuration is considered in this work. First, a multi-objective mixed-integer linear programming (MILP) model is formulated to maximize the reservation time in the storage area, and minimize the overstay time in the cleaning and inspecting areas. The relationship between constraints including flexible maintenance routes, train shunting conflicts, track occupation conflicts, and train arrival/departure times, are coordinated. Subsequently, a heuristic-based enhanced particle swarm optimization algorithm (EPSO) with two improvements is proposed to tackle this NP-hard problem. Specifically, three heuristic rules about the depth-first operation track allocation, the conflict-free bottleneck track allocation, and the right-shift track occupancy repair are designed to ensure the feasibility of the shunting schedule. Accordingly, a three-level decoding mechanism is designed to achieve a near-optimal shunting schedule with great train and route sequences. Two improvements on crossover and mutation operators are developed to enhance the exploration and exploitation ability. Finally, a real-world instance in China is solved to verify the effectiveness and efficiency of the proposed model and algorithm. Experimental results show that EPSO is relatively more effective than all the compared algorithms.

Automated summary

This paper proposes a solution to the EMU first-level maintenance shunting scheduling problem with flexible maintenance routes in a stub-end depot. By formulating a multi-objective MILP model and using a heuristic-based enhanced particle swarm optimization algorithm (EPSO), an approximate optimal schedule is achieved. Experimental results demonstrate the effectiveness and efficiency of the algorithm.