Optimization of Inbound and Outbound Vessel Scheduling in One-Way Channel Based on Reinforcement Learning DOI Creative Commons
Zhen Rong, Meng Sun, Qionglin Fang

et al.

Journal of Marine Science and Engineering, Journal Year: 2025, Volume and Issue: 13(2), P. 237 - 237

Published: Jan. 26, 2025

As the size and number of ships continue to grow, effective management vessel scheduling has become more important for efficient one-way channel port operation, whose characteristics significantly affect safety efficiency ports. This paper presents a reinforcement-learning-based approach optimize vessels in channel, aiming quickly identify solution that enhances operational efficiency. method models problem by incorporating navigational constraints, requirements, vessel-specific characteristics. Using Q-learning algorithm minimize wait times, it identifies an optimal solution. Experiments were conducted using real data from Dayao Bay Pier Dalian Port validate rationality effectiveness proposed model algorithm. The results show reinforcement learning achieved approximately 16% improvement quality compared genetic (GA) while requiring only half computation time. Additionally, reduced delay times over 40% relative traditional FCFS strategy, indicating superior overall performance. research efficient, intelligent scheduling, providing theoretical foundation further advancements this field enhancing decision support channels with practical implications.

Language: Английский

Optimization of Inbound and Outbound Vessel Scheduling in One-Way Channel Based on Reinforcement Learning DOI Creative Commons
Zhen Rong, Meng Sun, Qionglin Fang

et al.

Journal of Marine Science and Engineering, Journal Year: 2025, Volume and Issue: 13(2), P. 237 - 237

Published: Jan. 26, 2025

As the size and number of ships continue to grow, effective management vessel scheduling has become more important for efficient one-way channel port operation, whose characteristics significantly affect safety efficiency ports. This paper presents a reinforcement-learning-based approach optimize vessels in channel, aiming quickly identify solution that enhances operational efficiency. method models problem by incorporating navigational constraints, requirements, vessel-specific characteristics. Using Q-learning algorithm minimize wait times, it identifies an optimal solution. Experiments were conducted using real data from Dayao Bay Pier Dalian Port validate rationality effectiveness proposed model algorithm. The results show reinforcement learning achieved approximately 16% improvement quality compared genetic (GA) while requiring only half computation time. Additionally, reduced delay times over 40% relative traditional FCFS strategy, indicating superior overall performance. research efficient, intelligent scheduling, providing theoretical foundation further advancements this field enhancing decision support channels with practical implications.

Language: Английский

Citations

1