タブーサーチによる鉄道主要駅の整備作業ダイヤ自動作成手法の提案

抄録

Maintenance worker scheduling generates a daily schedule for each worker group to perform inspection and maintenance work such as cleaning rolling stock when superior trains turn around at a terminal station. Since the schedule varies from day to day according to changes in daily train timetables and rolling stock operations, it is required to be created for each day. This paper proposes a new method for automatic generation of maintenance worker schedules that can consider the priority of constraints by tabu search. It includes a new model and a generation method of an initial solution, and a generation method of neighborhood solutions. In order to confirm the effectiveness of the proposed method, we perform parameter sensitivity analysis of the tabu list. Moreover, we compare an actual schedule with schedules generated by the proposed method, a mathematical programming method that guarantees a theoretically optimal solution, and local search. The results indicate that the proposed method can generate a stable solution within a practical computation time of three minutes with reduced worker loads, regardless of the size of the train schedule.