Combining MAMBA and Attention-Based Neural Network for Electric Ground-Handling Vehicles Scheduling

To reduce airport operational costs and minimize environmental pollution, an increasing number of airports are transitioning from fuel-powered to electric ground-handling vehicles. However, the limited battery capacity of electric vehicles and the need for charging make the scheduling of these vehicles more complex. To address this scheduling problem, this paper proposes an electric ground-handling vehicle scheduling algorithm that combines the MAMBA model with an attention-based neural network. The MAMBA model is designed to process multi-dimensional features such as flight information, vehicle locations, service demands, and time window constraints. Subsequently, an attention mechanism-based neural network is developed to dynamically integrate vehicle states, service records, and operational and charging constraints, in order to select the most suitable flights for electric ground-handling vehicles to service. The experiments use flight data from Xiamen Gaoqi International Airport and compare the proposed method with CPLEX solvers, existing heuristic algorithms, and custom heuristic algorithms. The results demonstrate that the proposed method not only effectively solves the electric ground-handling vehicle scheduling problem and provides high-quality solutions, but also exhibits good scalability in different parameter settings and real-time scheduling scenarios.