仅测角机动目标跟踪原始对偶高斯粒子滤波  

Angle-only Maneuvering Target Tracking Using Primal-dual Gaussian Particle Filtering

作　　者：张宏伟 ZHANG Hongwei(School of Aeronautics and Astronautics,Sun Yat-sen University,Shenzhen 518107,China)

机构地区：[1]中山大学航空航天学院,深圳518107

出　　处：《电子与信息学报》2024年第4期1408-1417,共10页Journal of Electronics & Information Technology

基　　金：广东省基础与应用基础研究基金(2019A1515111099);中山大学青年培育项目(20lgpy72);中国科学院空间精密测量重点实验室开放基金(SPMT2021002,SPMT2022001)。

摘　　要：为消减仅测角机动目标跟踪系统中由时空不一致引起的投影基点偏移和高斯截断两类误差,该文采用映射表示和?1-?2,1稀疏正则表征时空因果一致约束,引入模糊综合贴近度建立次优建议分布,构建因果不变结构传递粒子集合以近似目标后验高斯积分,推导原始对偶高斯粒子滤波(PDGPF)算法。实验结果表明,相比交会测量最小二乘法,PDGPF算法定位旋翼无人机(UAV)的精度提升了18.4%~69.6%。相比于软约束辅助粒子滤波(SCAPF)算法,PDGPF算法在时空映射一致约束下能够自适应地修正粒子的权值,从而更为准确、稳定地跟踪机动点目标,整体计算负担减小了12.9%。To reduce the mapping basepoint offset and Gaussian truncation errors caused by spatiotemporalℓ1ℓ2,1inconsistency in angle-only maneuvering target tracking systems,mapping representation and-sparse regularization to represent spatiotemporal causal consistency constraints are used,the fuzzy comprehensive closeness is introduced to establish the suboptimal proposal distribution,the particle set in a causal invariant structure to approximate the Gaussian integration for target posterior is propagated,and the Primal-Dual Gaussian Particle Filtering(PDGPF)algorithm is derived.Simulation results show that,compared to the intersection measurement method with least squares,the accuracy for the PDGPF to locate a rotor Unmanned Aerial Vehicle(UAV)has improved by 18.4%~69.6%.Compared to the Soft Constrained Auxiliary Particle Filtering(SCAPF)algorithm,the PDGPF algorithm can adaptively correct the particle weights under the spatiotemporal mapping consistent constraints,obtaining more accurate and stable state estimation for tracking a maneuvering point target,reducing the overall computational burden by 12.9%.

关 键 词：机动目标跟踪 仅测角 时空不一致 原始对偶 因果不变 

分 类 号：TN953[电子电信—信号与信息处理]