基于Q-learning算法的SAR导引头制导律设计研究  

Research on shift handover design of terminal guidance law based on Q-learning algorithm

作　　者：贺鹏超 王宁 钱军 吴振华 刘曦[1] He Pengchao;Wang Ning;Qian Jun;Wu Zhenhua;Liu Xi(Xi'an Microelectronics Technology Institute,Xi'an 710065,China;Information Materials and Intelligent Sensing Laboratory of Anhui Province,Anhui University,Hefei 230601,China)

机构地区：[1]西安微电子技术研究所,西安710065 [2]信息材料与智能感知安徽省实验室,合肥230601

出　　处：《战术导弹技术》2023年第5期124-130,150,共8页Tactical Missile Technology

摘　　要：基于任务装订工作方式的传统空地SAR导引头存在模式固定、成像-打击交接班弹体约束高准确度要求、作战系统资源调配自由度低,无法满足雷达导引头智能化作战需求。针对该问题,提出基于全信息链条利用的强化学习智能制导律算法。该算法以雷达导引头当前位置和速度为输入状态,以导弹施加法向过载为动作空间,充分考虑到末制导弹载SAR成像视场角和前视跟踪打击脱靶量约束,以成像检测和跟踪打击脱靶量作为奖励函数。在连续的状态空间和动作空间中,雷达导引头能根据战场态势自主选择制导律动作。实验表明,所提算法可综合中制导单脉冲扫描、末制导聚束成像与前视跟踪打击弹体前倾角和曲线轨迹控制约束要求,并可根据目标成像结果合理调配SAR成像时间,提高打击目标识别准确率,具有一定的SAR导引头制导智能决策打击能力,并可灵活适应不同作战任务。Traditional air-ground synthetic aperture radar(SAR) seeker based on the task-binding mode faces the challenge of relatively fixed mode, the high accuracy requirement for imaging-strike missile body constraint, and low freedom in combat system resource allocation. To meet these challenges, an intelligent guidance algorithm with a complete information chain based on reinforcement learning technique is proposed.The algorithm takes the current position and speed of the radar seeker as state and the normal overload imposed by the missile as action, also takes fully considerations of the constraints of the borne SAR imaging field of view angle and forward looking tracking hit miss distance in the terminal guidance, and treat the detection rate and tracking hit miss distance as the reward function. In the continuous state space and action space, the radar seeker independently selects optimal action according to the battlefield situation. The simulation results show that the proposed guidance algorithm meets the requirements of the missile body rake angle and curve trajectory control in the entire midcourse guidance monopulse scanning, terminal guidance spotlight imaging and forward-looking tracking and hitting procedure. Moreover, the algorithm can flexiblely allocate SAR imaging time according to the target imaging results, enhance the accuracy of target recognition,and has the ability of intelligent decision-making and could flexibly adapt to different combat tasks.

关 键 词：SAR制导律 作战信息链条 弹体前置角 脱靶量 成像约束 智能决策 强化学习 

分 类 号：TJ765[兵器科学与技术—武器系统与运用工程]