基于强化学习的声诱饵航路规划方法  

An acoustic decoy route planning method based on reinforcement learning

作　　者：张旭 孔辉 查淞元 ZHANG Xu;KONG Hui;ZHA Songyuan(Shanghai Marine Electronic Equipment Research Institude,Shanghai 201108,China)

机构地区：[1]上海船舶电子设备研究所,上海201108

出　　处：《舰船科学技术》2025年第1期154-158,共5页Ship Science and Technology

摘　　要：本文给出一种基于强化学习的声诱饵航路规划方法。设计了适用于强化训练环境的步进式水声对抗仿真环境,通过该环境展示经典对抗态势与不利对抗态势。根据水声对抗的特点,设计了强化学习的观测空间、动作空间、奖励函数等关键要素。动作空间与奖励函数结合水声对抗特性进行了设计。借助Matlab平台进行深度神经网络的训练,并验证了训练结果,证明通过强化学习方法训练的声诱饵航路规划的有效性,具备将不利对抗态势转危为安的能力。In this paper,an acoustic decoy route planning method based on reinforcement learning is presented.A stepping underwater acoustic countermeasure simulation environment adapted to the intensive training environment is designed.Through this environment,the classical antagonistic situation and the adverse antagonistic situation are shown.According to the characteristics of underwater acoustic confrontation,the key elements of reinforcement learning such as observation space,action space and reward function are designed.The action space and reward function are designed in combination with underwater acoustic countermeasures.The deep neural network is trained by matlab platform,and the training results are verified,which proves the effectiveness of the acoustic decoy route planning trained by reinforcement learning method,and the ability of refusing to turn the adverse confrontation situation into safety.

关 键 词：强化学习 声诱饵 航路规划 

分 类 号：TP393.09[自动化与计算机技术—计算机应用技术]