基于DQN的雷达智能干扰决策方法

Radar Intelligent Jamming Decision Method Based on DQN

摘要:

对不同类型的雷达有源干扰进行了讨论，分析了不同干扰的作用机理，并对其干扰效果进行了仿真。讨论了深度Q-学习网络(deep Q-learning network，DQN)算法在传统算法基础上的改进，以及基于DQN的智能干扰决策流程，并通过仿真实验验证了基于 DQN的干扰决策算法的认知特性，同时测试了其在不稳定环境下的性能。仿真结果表明，采用基于DQN的干扰决策算法，能够使干扰机在缺乏先验知识的未知环境中，通过与环境的交互学习，不断提升干扰策略性能。

Abstract:

Different types of radar active jamming were discussed. The mechanism of different jamming was analyzed and the jamming effects were simulated. The improvement of deep Q-learning network (DQN) algorithm compared with the traditional algorithm and the intelligent jamming decision process based on DQN were discussed. The cognitive characteristics of jamming decision algorithm based on DQN were verified by simulation experiments, and the performance of the algorithm in unstable environment was tested. The simulation results show that the jamming decision algorithm based on DQN can continuously improve the interference strategy of jammer through interactive learning with the environment in the unknown environment without prior knowledge.