Discrete Dynamic Berth Allocation Optimization in Container Terminal Based on Deep Q-Network

Mathematics, Journal Year: 2024, Volume and Issue: 12(23), P. 3742 - 3742

Published: Nov. 28, 2024

Effective berth allocation in container terminals is crucial for optimizing port operations, given the limited space and increasing volume of traffic. This study addresses discrete dynamic problem (DDBAP) under uncertain ship arrival times varying load capacities. A novel deep Q-network (DQN)-based model proposed, leveraging a custom state space, rule-based actions, an optimized reward function to dynamically allocate berths schedule vessel arrivals. Comparative experiments were conducted with traditional algorithms, including ant colony optimization (ACO), parallel (PACO), combined genetic algorithm (ACOGA). The results show that DQN outperforms these methods significantly, achieving superior efficiency effectiveness, particularly high variability arrivals conditions. Specifically, reduced total waiting time vessels by 58.3% compared ACO (262.85 h), 57.9% PACO (259.5 57.4% ACOGA (257.4 109.45 h. Despite its impressive performance, requires substantial computational power during training phase sensitive data quality. These findings underscore potential reinforcement learning optimize Future work will explore multi-agent (MARL) real-time adaptive mechanisms further enhance robustness scalability model.

Language: Английский

Synchronized optimization of the logistics system of a tobacco high-bay warehouse under production task fluctuations

Engineering Research Express, Journal Year: 2024, Volume and Issue: 6(4), P. 045580 - 045580

Published: Dec. 1, 2024

Abstract Cigarette companies often utilize a logistics system where high-bay warehouse areas for semi-finished and finished tobacco products are located close together linked via trolley tracks. However, fluctuations in production tasks, such as adding new handling emergency orders, canceling existing can introduce challenges to production, including high storage capacity ratios limited transportation capacity. Therefore, this study proposes synchronized optimization method the warehouses designed improve efficiency. A allocation model is developed balance between two tackle ratios. The efficiency improved by minimizing stacker time optimizing shelf center of gravity, with warehouses’ serving key constraints. neural network introduced enhance predicting optimal number trolleys needed each functional area based on task volume, boosting overall Experimental analysis demonstrates that proposed coordinated operation enhances 36.82% 8%. This achieves cigarette enterprises designing shared coordination scheme locations trolleys. Lastly, effectively supports industry addressing dynamic demands provides valuable insights practical engineering applications.

Language: Английский