| 摘要: |
| 针对多雷达系统多目标跟踪资源分配问题，建立了综合考虑目标威胁度与跟踪精度的成本函数，以雷达驻留时间为控制量，构建了有约束的部分可观测马尔可夫决策过程资源分配模型，采用拉格朗日松弛法将优化问题分解为多个无约束子问题，使用预演算法实现非短视近似最优求解。依据多雷达信息融合架构差异，分别给出集中式和分布式两种实现方案，并通过静态场景和动态场景完成算法验证。仿真结果表明，本文所提算法可实现多雷达资源的一体化非短视快速调度，并能按威胁等级差异化分配跟踪资源。 |
| 关键词: 目标威胁度 跟踪精度 多雷达 非短视；资源分配 |
| DOI： |
| 分类号:TN95 |
| 基金项目:国家自然科学基金项目面上项目，湖北省自然科学基金面上项目 |
|
| Multi-radar resource allocation algorithm for joint optimization of target threat degree and tracking accuracy |
|
胡旭超, 曲智国, 刘维建, 谭贤四
|
|
|
| Abstract: |
| Aiming at the resource allocation problem for multi-target tracking in multi-radar systems, a cost function that comprehensively considers target threat degree and tracking accuracy is established. With radar dwell time as the control variable, a constrained resource allocation model based on the Partially Observable Markov Decision Process (POMDP) is constructed. The Lagrangian relaxation method is adopted to decompose the optimization problem into multiple unconstrained subproblems, and the Rollout algorithm is used to achieve a non-myopic approximate optimal solution. According to the differences in multi-radar information fusion architectures, two implementation schemes, namely centralized and distributed, are proposed respectively, and the algorithm is verified through static and dynamic scenarios. Simulation results demonstrate that the proposed algorithm can realize the integrated non-myopic rapid scheduling of multi-radar resources and allocate tracking resources differentially according to threat levels. |
| Key words: target threat degree tracking accuracy multi-radar non-myopic resource allocation |