Collaborative optimization of truck scheduling in container terminals using graph theory and DDQN

Area of Science:

• Operations Research
• Artificial Intelligence
• Logistics Management

Background:

• Container terminals are critical hubs in global trade and logistics.
• Inefficient truck scheduling leads to low truck utilization and long equipment waiting times, hindering port operations.

Purpose of the Study:

• To develop an optimized truck scheduling strategy for container terminals.
• To minimize the maximum completion time of terminal equipment and enhance overall efficiency.

Main Methods:

• A container terminal simulation model based on graph theory was developed.
• A collaborative scheduling algorithm for truck fleets using Deep Double Q-Networks (DDQN) was proposed.
• The DDQN algorithm incorporated five heuristic rules, refined state features, and reward functions.

Main Results:

• The DDQN algorithm consistently identified optimal scheduling strategies.
• It outperformed existing heuristic rules and the Deep Q-Network (DQN) algorithm.
• Significant reductions in quay crane waiting times and equipment completion times were observed, alongside improved truck utilization.

Conclusions:

• The proposed DDQN-based algorithm effectively optimizes truck scheduling in container terminals.
• This leads to substantial improvements in port efficiency and resource utilization.
• The approach offers a promising solution for enhancing container terminal operations.