针对多传感器多目标跟踪中存在的目标漏检和杂波干扰问题,提出了一种基于随机有限多目标跟踪理论的传感器调度方法。首先依据部分可观马尔科夫决策过程和随机有限集建立调度模型,对多目标运动状态和传感器辐射代价进行数学描述,并基于跟踪精度和辐射代价的平衡创建优化函数。随后,采取高斯混合概率假设密度平滑滤波算法估计目标 状态,以最优子模式分配距离为指标,实现多目标长时跟踪精度的预测。使用辐射度影响指标量化辐射代价,设置辐射告警作用距离限制代价累积,并利用隐马尔科夫模型滤波算法实现长时辐射代价预测。最后,以长时跟踪精度和辐射代价的预测为决策依据,结合最小工作时长,制定了传感器调度方法及实现流程。通过仿真验证,所提调度方法在多指标平衡优化方面具有明显优势,实现了杂波环境下多传感器多目标跟踪的合理调度。

For the issue of target missing and clutter interference in multi-sensor multi-target tracking, a scheduling method for sensors based on random finite set multi-target tracking theory is proposed. Firstly, the scheduling model is established based on partially observable Markov decision process and random finite set, the multi-target motion state and sensor radiation cost are described mathematically, and the objective optimization function is created based on the balance of tracking accuracy and radiation cost. Then, the Gaussian mixture probability hypothesis density smoothing filter is used to estimate the target state, and the optimal subpattern assignment distance is used as the index to realize the prediction of multi-target non-myopic tracking accuracy. The emission level impact is used to quantify the radiation cost, the radiation warning distance is set to limit the accumulation of costs,and the hidden Markov model filter is used to predict the non-myopic radiation cost. Finally, based on the prediction of non-myopic tracking accuracy and radiation cost, the sensor scheduling method is formulated with the minimum timestep constraint. Simulation results show that the proposed scheduling method has obvious advantages in multi-index balanced optimization, and realizes the reasonable scheduling of multi-sensor multi-target tracking in clutter environment.

TN957. 51

中国国防科技预研资助项目(41404030101,41404030102)