多传感器混合多模型估计的误差互相关性及其融合算法研究  

On Error Cross-correlation and Fusion Algorithm for Multi-sensor Hybrid Multiple Model Estimation

作　　者：乔向东[1] 李涛[2] 杨仝[1] 李鸿艳[1] 

机构地区：[1]空军工程大学电讯工程学院,陕西西安710077 [2]中国电子科技集团第28研究所,江苏南京210007

出　　处：《电子学报》2010年第4期804-810,共7页Acta Electronica Sinica

基　　金：国家自然科学基金(No.60774091);陕西省自然基金(No.2007-F24)

摘　　要：对局部节点状态估计间误差相关性的处理是分布式估计融合或航迹融合的关键要素;针对当前分布式融合理论中关于混合多模型估计融合研究的空白,首先推导得出了采用相同模型成分的各局部节点交互多模型状态估计的误差互协方差矩阵的递推计算方法;其次,讨论了所得非对称实误差互协方差矩阵的正定特性,并分析了此类误差相关性与混合多模型估计算法中模型过程噪声之间的变化关系;上述结果使得基于互协方差组合融合算法的交互多模型状态估计融合成为可能,仿真实验亦验证了其有效性,相对其它不考虑误差相关性的融合算法,融合结果也更为真实.Treatment of error correlation between local estimations is principal part of distributed estimation fusion and track fusion.In view of current situation,in which fusion of multi-sensor hybrid multiple model estimations is still not touched,at first,recursive computation methods for error cross-covariance matrix between local estimations obtained by interacting multiple model estimators whose model set are completely same or partly same is derived.Second,through simulation,the positive definite of cross-covariance matrix is discussed,and its relationship with process noise of maneuver motion model of IMM estimator is investigated.Further,above achievement makes applying Barshalom-Campo algorithm to fusion of IMM estimations as possible,its validity is verified through Monte-Carlo simulation.Compared with other fusion algorithms that ignore error correlation,performance of BC algorithm is more truly.

关 键 词：分布式融合 误差相关性 互协方差 混合估计 交互多模型 

分 类 号：TN953[电子电信—信号与信息处理]