多目标环境下基于分布式融合思想的误差估计方法

Algorithm for Bias Estimation Based on Distribution Fusion in Multi-target Environment

针对多雷达多目标跟踪过程中分布未知的系统误差估计问题,提出了基于"分布式融合思想"的误差估计方法。给出相应误差估计方法的计算公式,利用改进截断奇异值方法来减轻矩阵病态性的影响,提高误差估计的稳健性。设置了两种不同的系统误差仿真场景,对"分布式"误差估计方法在两种情形下的估计性能进行了仔细对比分析。结合"分布式"误差估计方法与"集中式估计"方法所体现出的优缺点,提出了一种将两种方法结合起来的系统误差估计算法,算法通过合理选择阈值门限η,能够在多雷达多目标且系统误差分布未知的复杂环境下对两种误差估计算法自适应地进行切换,从而充分发挥两种误差估计算法各自的优点,给出更好的误差估计结果。

For unknown distribution system error estimation in multi-sensor and multi-target tracking systems, an algorithm for bias estimation based on distribution fusion is proposed. Errors computing formula, ill-condition of matrix is alleviated and estimation results becomes more robust by using TSVD （ truncated singular value decomposition） method. For the performance and character of this methods in different system error distribution simulation environment, a new algorithm combined this method with centralized estimation for solving unknown distribution system error estimation in multi-radar and multi-target tracking systems is proposed. By choosing a proper threshold, this algorithm can adaptive shear in two methods. This algorithm can make full use of each method＇s advantage, then give a better estimation results. And the estimation method module can be expanded to fit for adaptive estimation in a more complex errors distribution environment.