| 基于强化学习的锚泊辅助动力定位系统智能决策研究 |
| |
| 引用本文： | 余尚禹,王磊,李博,衣凡.基于强化学习的锚泊辅助动力定位系统智能决策研究[J].海洋工程,2019,37(6):49-61. |
| |
| 作者姓名： | 余尚禹 王磊 李博 衣凡 |
| |
| 作者单位： | 上海交通大学 海洋工程国家重点实验室，上海 200240,上海交通大学 海洋工程国家重点实验室，上海 200240；
高新船舶和深海开发协同创新中心，上海 200240,上海交通大学 海洋工程国家重点实验室，上海 200240,上海交通大学 海洋工程国家重点实验室，上海 200240 |
| |
| 基金项目： | 工信部课题“第七代半潜平台模型试验及技术研究”系泊定位技术（［2016］546）；国家重点研发计划（2016YFC0303405） |
| |
| 摘 要： | 针对半潜平台锚泊辅助动力定位系统的最优定位点问题,设计了基于强化学习中深度神经网络的Q学习(DQN)控制策略的锚泊辅助动力定位的智能决策系统。该决策系统中DQN方法与比例—积分—微分(PID)控制方法相结合使用,实现系统优化。在基于机器人操作系统(ROS)平台的动力定位时域模拟程序中进行数值仿真,仿真结果验证了该系统在定位点决策问题上的可靠性和有效性,从而使半潜平台在面对未知海况时,均能寻找到最优定位点,在保证锚泊辅助动力定位系统可靠性的同时降低功率消耗,提高经济性。
|
| 关 键 词： | 动力定位系统 强化学习 人工神经网络 设点定位 |
Intelligent setpoint control of thruster-assisted postion mooring of a semi-submersible platform based on reinforcement learning |
| |
| YU Shangyu,WANG Lei,LI Bo and YI Fan.Intelligent setpoint control of thruster-assisted postion mooring of a semi-submersible platform based on reinforcement learning[J].Ocean Engineering,2019,37(6):49-61. |
| |
| Authors: | YU Shangyu WANG Lei LI Bo and YI Fan |
| |
| Institution: | State Key Laboratory of Ocean Engineering, Shanghai Jiao Tong University, Shanghai 200240, China,State Key Laboratory of Ocean Engineering, Shanghai Jiao Tong University, Shanghai 200240, China;
Collaborative Innovation Center for Advanced Ship and Deep-Sea Exploration, Shanghai 200240, China,State Key Laboratory of Ocean Engineering, Shanghai Jiao Tong University, Shanghai 200240, China and State Key Laboratory of Ocean Engineering, Shanghai Jiao Tong University, Shanghai 200240, China |
| |
| Abstract: | To improve the working performance of the thruster-assisted position mooring (PM) system, an intelligent setpoint decision strategy based on reinforcement learning is proposed. The Deep Q network (DQN) is used in combination with the PID control to achieve the optimization of the PM system. The simulation results based on ROS platform show that when operating in moderate sea conditions, the proposed intelligent setpoint decision strategy can successfully identify the optimal setpoint at which the power consumption of the thrusters can keep at a relatively low level. |
| |
| Keywords: | positioning mooring system reinforcement learning artificial neural networks optimal setpoint |
| 本文献已被 CNKI 等数据库收录！ |
| 点击此处可从《海洋工程》浏览原始摘要信息 |
| 点击此处可从《海洋工程》下载免费的PDF全文 |
|
