摘要
当检测环境中有一个统一的噪声源时,各个传感器的测量噪声是相关的.许多算法都假设各个传感器的测量噪声是不相关的,因为这是保证测量噪声方差矩阵平行解耦的必要条件.基于矩阵相似变换的理论,提出了一种使多传感器相关测量噪声方差矩阵解耦的方法,该方法使多传感器数据融合测量模型转化成测量噪声不相关的新模型,推导了存在相关测量噪声的多传感器数据融合最优状态估计算法.当测量噪声不相关时,算法与以往的具有不相关噪声的最优算法相同.
Measurement noises are correlated when the same noise source. Many fusion estimation algorithms assumed that measurement noise processes between the sensors are uncorrelated, which is the critical sufficient condition that allows successful parallel decomposition of the covariance matrix of the measurement noise. The matrix resemble transform theory was successfully used for parallel decomposition of the covariance matrix of correlated measurement noise and transform the linear observation models into new observation models. The optimal data fusion estimation algorithms are presented. When measurement noises are uncorrelated, the algorithms are equivalent to the optimal algorithms with uncorrelated measurement noise.
出处
《浙江大学学报（工学版）》
EI
CAS
CSCD
北大核心
2003年第1期60-64,共5页
Journal of Zhejiang University：Engineering Science
