Integrated optimizing of double cantilever yard cranes scheduling and vehicle parking positions in U-shaped container terminal

The U-shaped automated container terminal improves traffic flow and reduces congestion by physically separating external trucks from intelligent guided vehicle (IGV) transport channels. However, at the scheduling level, there exists a reciprocal influence among IGVs, collection trucks, and yard cranes. This study addresses and resolves issues related to yard crane scheduling and vehicle parking position selection. A mixed-integer programming model is developed to minimize completion time, and a reinforcement learning-based genetic simulated annealing algorithm is designed to solve the model. The algorithm applies elitism, crossover, and mutation operations, uses an annealing mechanism to avoid local optima, and introduces a reinforcement learning-based decision-making mechanism to dynamically adjust relevant parameters and guide the algorithm’s decisions during the search process. The effectiveness of the proposed method and algorithm is validated through numerical experiments, demonstrating an average yard crane completion time reduction of 9.06% and a maximum reduction of 11.98%.