主瓣干扰是雷达面临的重要威胁,通过多雷达组网协同发挥体系效能是对抗此类威胁的有效手段。文中针对主瓣干扰下雷达组网协同探测系统跟踪性能下降的问题,提出一种基于强化学习的多目标跟踪主被动联合的资源调度方法。首先,给出干扰条件下的雷达主动量测模型和被动定位模型;然后,利用雷达在主动和被动工作模式下精度随干扰强度的不同变化趋势,设计了以最小化跟踪误差为目标的评价函数;最后,使用近端策略优化算法的强化学习智能体进行雷达主被动工作模式选择,将剩余的多雷达驻留时间分配作为凸优化问题进行求解。仿真实验结果表明,该方法相比传统的雷达工作模式分配策略,能够提升雷达组网协同探测系统在强干扰环境下的跟踪性能,从而提高系统抗干扰能力。

Mainlobe jamming is an essential threat to radars. Resource management collaborating multiple radars is an effective strategy to counter such threats. In this paper, a reinforcement learning-based technique of dynamically selecting multiple radar active and passive work modes is proposed to improve the overall performance of multi-target tracking under mainlobe jamming. The active measurement and passive localization models in jamming conditions are presented. An objective function to minimize tracking error is designed by utilizing the opposite correlation between precision and jamming intensity of the two work modes. Finally, a reinforcement learning agent employing the proximal policy optimization algorithm is used to select the radars′ work modes. The remaining variables of multi-radar dwell time are then optimized as a convex optimization problem. Simulation results demonstrate that, compared to per-determined work mode selection strategies, the proposed approach improves the tracking performance of the radar network in jamming environments, thereby improving the anti-jamming capability.

TN957

国家自然科学基金资助项目(62301295);清华大学自主科研计划(20234180184)