基于部分可观察马尔可夫决策过程的多被动传感器组网协同反隐身探测任务规划  

Mission Planning of Passive Networked Sensors for Cooperative Anti-stealth Detection Based on POMDP

作　　者：万开方[1] 高晓光[1] 李波[1] 梅军峰[1] 

机构地区：[1]西北工业大学电子信息学院,陕西西安710072

出　　处：《兵工学报》2015年第4期731-743,共13页Acta Armamentarii

基　　金：全国高等院校博士学科点专项科研基金项目(20116102110026);空军装备预先研究项目(402040103);西北工业大学基础研究基金项目(JC201144);西北工业大学科研启动基金项目(GAKY300701)

摘　　要：针对反隐身作战需求,提出多被动传感器组网协同战术。为提升反隐身探测效能,引入部分可观察马尔可夫决策过程(POMDP)理论,分析了POMDP任务规划要素,建立起多被动传感器组网协同反隐身探测任务规划POMDP模型。建立了多被动传感器协同控制系统结构,提出了基于无迹卡尔曼滤波(UKF)的信念状态更新方法和基于蒙特卡洛Rollout采样(MCRS)的Q值估计方法,并设计了CCSP基本策略。仿真结果表明,所建立的模型能够实现多被动传感器的高效管理调度,能够控制多被动传感器对隐身目标进行有效探测跟踪,即模型有效性得到了验证。A cooperative strategy for multiple passive networked sensors to satisfy the requirements of anti-stealth is proposed. A POMDP model for the multiple passive networked sensors is established to improve the effectiveness of the anti-stealth detection. The elements of POMDP model are analyzed compre- hensively based on the POMDP theory. A cooperative control system of multi-passive-sensors is put forward, and a UKF method for belief state updating together with a MCRS-based （Monte-Carlo Rollout sampling） algorithm for Q-value estimation is presented. A basic policy of CCSP is designed. Experiments show that the proposed model can be used for the efficient management and scheduling of multi- passive-sensors, as well as the effective detection and tracking of the stealth targets by deploying these sensors. The validity of the proposed algorithm is demonstrated.

关 键 词：控制科学与技术 传感器技术 反隐身 多传感器组网 部分可观察马尔可夫决策过程 信念状态 任务规划 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]