Abstract: Aiming at the issue of multi-target tracking in a clutter environment,a non-myopic scheduling method of sensor radiation control is proposed to improve the tracking accuracy and reduce the usage cost.Based on the partially observable Markov decision process (POMDP) and random finite set (RFS),the target state model and observation model are established;according to the idea of Interception Probability,an improved radiation risk quantification method is proposed;for balancing the tracking accuracy and radiation risk effectively,an optimization function is created.The Gaussian mixture probability hypothesis density (GM-PHD) smoothing filter is used to update the multi-target state,and optimal subpattern assignment (OSPA) distance is used as the measurement index to propose the prediction method of multi-target non-myopic tracking accuracy;combined with the maximum range of target radiation warning,the prediction method of non-myopic radiation cost is proposed;and the sensor scheduling process is designed.The feasibility and effectiveness of the scheduling model and method are verified by simulation experiments.The results show that using the improved radiation risk quantification method for scheduling can further improve the utilization efficiency of sensor resources on the basis of realizing radiation control.The proposed scheduling method can ensure the accuracy of multi-target tracking,effectively reduce the radiation risk and switching times of sensors,and realize the reasonable scheduling of multi-sensor multi-target tracking in clutter environment.

Key words: sensor scheduling, random finite set, radiation risk, interception probability, target tracking, clutter environment

摘要： 针对杂波环境下的多目标跟踪问题,为提高跟踪精度、降低使用代价,提出了一种传感器辐射控制的长时调度方法。首先,基于部分可观马尔可夫决策过程(partially observable Markov decision process,POMDP)和随机有限集(random finite set,RFS)建立目标状态模型和观测模型;基于截获概率的思想,提出了改进的辐射风险量化方法;并以跟踪精度和辐射风险之间的有效平衡为目标,创建了优化函数。其次,采取高斯混合概率假设密度(Gaussian mixture probability hypothesis density,GM-PHD)平滑滤波器更新多目标状态,以最优子模式分配(optimal subpattern assignment,OSPA)距离为衡量指标,提出了多目标长时跟踪精度的预测方法;结合目标的辐射告警最大作用距离,提出了长时辐射代价的预测方法,设计了传感器调度流程。最后,通过仿真实验,验证了调度模型、调度方法的可行性。结果表明,使用改进的辐射风险量化方法进行调度,可以在实现辐射控制的基础上,进一步提高传感器资源的利用效率;所提调度方法在确保多目标跟踪精度的同时,有效降低了传感器辐射风险和切换次数,实现了杂波环境下多传感器多目标跟踪的合理调度。

关键词: 传感器调度, 随机有限集, 辐射风险, 截获概率, 目标跟踪, 杂波环境