基于强化学习的成品油船装载方案自主生成技术研究
Reinforcement learning-based autonomous generation technology for product oil tanker loading schemes
尼洪涛 1周清基 2柴松 3齐鸣4
作者信息
- 1. 苏州城市学院 计算科学与人工智能学院,江苏 苏州 215104
- 2. 天津大学 海洋科学与技术学院,天津 300072
- 3. 苏州载诺信息科技有限公司,江苏 苏州 215008
- 4. 上海中船船舶设计技术国家工程研究中心有限公司,上海 201114
- 折叠
摘要
[目的]旨在基于强化学习方法研究液货舱装载方案自主生成技术.[方法]以实际运营的成品油船载货量作为输入,以货舱及压载舱的装载率为目标,基于Unity ML-Agents构建智能体与环境,通过PyTorch框架对智能体进行训练,提出一种综合考虑装载时间与纵倾变化幅度的奖励函数计算方法,并以算例分析来验证所提方法的有效性.[结果]结果显示,所训练的智能体能够学习良好的策略,并实现液货舱装载方案的自主生成.[结论]研究结果表明,将强化学习用于解决多目标条件下的液货舱装载方案自主生成是合理可行的.
Abstract
[Objective]This paper focuses on using reinforcement learning-based automatic generation tech-nology to generate loading and unloading schemes for the liquid cargo tanks of oil tankers.[Methods]Us-ing the cargo capacity of an actual operating oil tanker as the input and the loading rates of the cargo tank and ballast water tank as the targets,an intelligent agent and environment are built based on Unity ML-Agents.The agent is trained using the PyTorch framework,and a reward function calculation method that comprehensively considers the loading time and changes in the trim amplitude is proposed.Finally,example analysis is carried out to validate the feasibility of the proposed method.[Results]The results show that the trained agent can learn effective strategies for achieving the autonomous generation of liquid cargo tank loading schemes.[Conclusions]This study proves that it is reasonable and feasible to apply reinforcement learning to solve the problem of the autonomous generation of liquid cargo tank loading schemes under multi-objective condi-tions.
关键词
自动化装卸/液货舱/机器学习/方案优化Key words
automatic loading and unloading/liquid cargo tank/machine learning/scheme optimization引用本文复制引用
基金项目
天津市交通运输科技发展计划(G2022-48)
出版年
2024
国家科技期刊平台
NSTL
万方数据