Disturbance-Aware Closed-Loop Rescheduling for Airport Ground Operations Via LLM-Enhanced Decision Support

Airport ground operation scheduling is challenging due to strong coupling among aircraft, resources, and spatial constraints, as well as frequent disruptions. Traditional methods either assume deterministic environments or rely on full rescheduling, leading to high computational cost and poor schedule stability. To address these issues, this paper proposes a disturbance-aware closed-loop rescheduling framework integrating disturbance perception, impact evaluation, and adaptive scheduling. A selective rescheduling mechanism is developed to classify disturbances by impact level, while a partial schedule freezing strategy preserves unaffected tasks and limits adjustment scope within a time-window scheme. In addition, a large language model (LLM) is introduced as a semantic interface to transform unstructured disturbance information into structured inputs for scheduling. A hierarchical heuristic scheduling method combined with simulation is adopted to efficiently generate feasible solutions. Experimental results demonstrate that the proposed framework can effectively resolve conflicts and produce stable rescheduling plans with improved computational efficiency, providing a practical and scalable solution for airport ground operations under dynamic environments.