航空学报 > 2014, Vol. 35 Issue (8): 2299-2306   doi: 10.7527/S1000-6893.2014.0076

基于MCMC-Gibbs采样的天波超视距雷达联合状态估计与模式辨识

冯肖雪, 梁彦, 焦连猛   

  1. 西北工业大学 自动化学院, 陕西 西安 710072
  • 收稿日期:2013-11-28 修回日期:2014-04-25 出版日期:2014-08-25 发布日期:2014-05-05
  • 通讯作者: 梁彦,Tel.:029-88431308,E-mail:liangyan@nwpu.edu.cn E-mail:liangyan@nwpu.edu.cn
  • 作者简介:冯肖雪女,博士研究生。主要研究方向:估计理论,信息融合,目标跟踪与识别。Tel:029-88431306,E-mail:fwjiangyan@163.com;梁彦男,博士,教授,博士生导师。主要研究方向:估计理论,信息融合,目标跟踪与识别。Tel:029-88431308,E-mail:liangyan@nwpu.edu.cn
  • 基金资助:

    国家自然科学基金(61135001,61074179,61374023,61374159);西北工业大学博士论文创新基金(CX201320)

Joint State Estimation and Mode Identification based on MCMC-Gibbs Sampling for OTHR

FENG Xiaoxue, LIANG Yan, JIAO Lianmeng   

  1. School of Automation, Northwestern Polytechnical University, Xi'an 710072, China
  • Received:2013-11-28 Revised:2014-04-25 Online:2014-08-25 Published:2014-05-05
  • Supported by:

    National Natural Science Foundation of China (61135001, 61074179, 61374023, 61374159); Doctorate Foundation of the Northwestern Polytechnical University (CX201320)

摘要:

天波超视距雷达(OTHR)目标跟踪面临着“三低”(低检测概率、低数据率、低测量精度)和“多路径”(多条传播路径)的严峻挑战,准确的传播模式辨识与精确的目标状态估计是改善跟踪能力的关键。针对上述问题,提出了一种基于马尔科夫蒙特卡洛吉布斯(MCMC-Gibbs)采样的OTHR联合状态估计与模式辨识算法,该算法通过MCMC-Gibbs采样求取当次迭代当前拍最优的关联矩阵,进而利用同时多量测滤波进行状态和协方差更新,最后引入联合估计与辨识风险函数寻求最优的模式辨识与状态估计结果。不同仿真参数下仿真结果表明该算法的有效性,同时该算法在径向距和方位角估计精度上均高于多路径概率数据关联算法(MPDA),但这是以计算量为代价的。

关键词: 天波超视距雷达, 模式辨识, 状态估计, 马尔科夫蒙特卡洛吉布斯采样, 联合估计与辨识贝叶斯风险

Abstract:

Target tracking of over the horizon radar (OTHR) faces the challenge of the low detection probability, low sampling rate, low measurement accuracy and the multipath propagation. Both mode recognition of multipath propagation and state estimation significantly affect the tracking performance. In this paper, the method of joint state estimation and mode identification based on Markov Chain Monte Carlo-Gibbs (MCMC-Gibbs) sampling for OTHR target tracking is proposed. Validation gates are firstly constructed for every mode to generate only those hypotheses that satisfy the validation gate requirement to eliminate the number of hypotheses significantly. Then the association events are obtained through MCMC-Gibbs sampling to further calculate the decision cost. Next, multiple simultaneous measurement filters are proposed to update the conditional state estimation and covariance for estimation cost. Finally, Bayes risk for joint decision and estimation is introduced to find the optimal solution. Simulation results show the effectiveness of the proposed method compared with the multipath data association tracker (MPDA) method at some sacrifice to computation cost.

Key words: over the horizon radar, pattern identification, state estimation, MCMC-Gibbs sampling, Bayes risk for joint decision and estimation

中图分类号: