基于DDPG算法的智能水雷协同攻击方法
Coordinated Attack Method for Intelligent Mine Based on DDPG Algorithm
蒋平 1李加强 2彭荆明2
作者信息
- 1. 西北工业大学航海学院 西安 710114;宜昌测试技术研究所 宜昌 443000
- 2. 宜昌测试技术研究所 宜昌 443000
- 折叠
摘要
智能化是近年来水雷技术发展的重点方向,而具备协同攻击能力是水雷集群智能化的典型特征.论文针对水雷作战中的典型任务场景,基于深度确定性策略梯度网络(DDPG)的强化学习算法,设计了实时和终端两种奖励函数,提高了算法的训练有效性,并通过仿真实验验证.实验结果表明,智能水雷集群在完成训练后,能够较好地执行协同截击、围攻、拦截等攻击任务,可有效提升水雷作战效能.
Abstract
Intelligence is the key direction of mine technology development in recent years,and capability of coordinated attack is a typical characteristics of mine swarm intelligence.Aiming at typical mission in mining capability,this paper designs two reward functions of real-time and terminal based on DDPG,which improves the training effectiveness of the algorithm and is verified by simulation experiments.The experiment results show that the intelligent mine swarm after the training can better perform the cooperative interception,siege,interception and other attack tasks,and effectively improve the operational efficiency of the mine.
关键词
智能水雷/协同攻击/深度强化学习/策略梯度Key words
intelligent mine/coordinated attack/deep reinforcement learning/policy gradients引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据