基于强化学习的多机协同传感器管理

网络化战争中,机载雷达在实现对目标信息持续获取的同时保证载机安全生存是亟待解决的问题。对此,以多机协同作战安全转场任务为背景,提出基于深度强化学习算法的智能传感器管理方法。首先,综合考虑信号辐射量与目标威胁因素,计算目标运动过程中的实时威胁隶属度。其次,在强化学习框架下对雷达-目标分派问题建模,利用神经网络逼近动作-值函数,并根据时序差分算法进行参数更新。仿真结果表明,相比于传统调度方法,所提算法有效提升了任务成功率,缩短了任务完成用时。

Multi-airborne cooperative sensor management based on reinforcement learning

In the networked war, it is urgent that airborne radar can continuously acquire target information while ensuring the safe survival. Focusing on this problem, in the context of safe transition tasks of multi-airborne cooperative operations, this paper proposes a intelligent sensor management method based on deep reinforcement learning. First, the real-time threat membership is calculated considering the signal radiation and several threat factors. Then, the radar-target assignment problem is modeled in a reinforcement learning framework. The neural network is used to approximate the action-value function, and the parameters are updated according to the temporal-difference algorithm. It can be seen from the simulation that the proposed algorithm improves the task success rate and shortens the time of task completion compared with the traditional scheduling methods.