一种基于强化学习的绞吸挖泥船施工参数智能自主寻优方法研究  

Research on an intelligent self-optimization method for cutter suction dredger construction parameters based on reinforcement learning

作　　者：鲁嘉俊 杨波 徐婷 LU Jia-jun;YANG Bo;XU Ting(CCCC National Engineering Research Center of Dredging Technology and Equipment Co.,Ltd.,Shanghai 201208,China)

机构地区：[1]中交疏浚技术装备国家工程研究中心有限公司,上海201208

出　　处：《中国港湾建设》2022年第8期71-75,共5页China Harbour Engineering

摘　　要：针对现阶段绞吸船疏浚控制系统需要智能优化疏浚生产效率的需求,提出了一种基于强化学习的绞吸挖泥船施工参数智能自主寻优方法,首先采用信息增益率的方法挑选施工过程的控制变量,组成多元的训练数据组,然后搭建包含连续动作空间、状态转移和奖惩函数的强化学习环境模型;智能体根据算法给出的随机动作执行指令并反馈状态信息,通过与环境的交互学习逐渐获得最优策略,实现绞吸船疏浚参数的自主寻优。利用实船采集的数据进行仿真实验,结果表明基于强化学习的疏浚参数自主寻优方法能在不确定环境条件下快速有效地学习和达到目标,证明了此方法的合理性和有效性。In response to the current need for intelligent optimization of dredging production efficiency in the cutter suction dredger dredging control system, an intelligent self-optimization method for the construction parameters of a cutter suction dredger based on reinforcement learning was proposed. Firstly, the control variables in the construction process were selected by using the method of information gain rate to form a multiple training data set, and then a reinforcement learning environment model that includes continuous action space, state transition, and reward and punishment functions was established. The agent executes instructions and feedback state information according to the random actions given by the algorithm, gradually obtain the optimal strategy through interactive learning with the environment, and realize the self-optimization of dredging parameters of the cutter suction dredger. Simulation experiments were carried out using data collected from real ships, and the results show that the self-optimization method of dredging parameters based on reinforcement learning can quickly and effectively learn and achieve goals under uncertain environmental conditions, which proves the rationality and effectiveness of this method.

关 键 词：强化学习 绞吸挖泥船 疏浚作业 自主寻优 

分 类 号：U615.351[交通运输工程—船舶及航道工程] U615.38[交通运输工程—船舶与海洋工程]