针对多传感器数据融合中互协方差未知的问题,提出了一种基于分离协方差交叉的全局反馈航迹融合方法,将误差协方差矩阵分离为相关信息误差协方差矩阵和独立信息误差协方差矩阵.在此基础上,提出了分离形式的卡尔曼滤波算法,并设计了全局信息反馈的分布式融合结构,克服了简单凸组合融合算法和协方差交叉算法中将所有的信息一起考虑的缺陷.仿真实验表明:本文算法比简单凸组合融合算法和协方差交叉融合方法具有更小的均方误差. In order to solve the problem of unknown covariance variance in multi-sensor data fusion, a global feedback trajectory fusion method based on cross-covariance crossover is proposed. The covariance matrix is ​​separated into covariance matrix of correlated information error and covariance matrix of independent information error Based on this, a new Kalman filtering algorithm is proposed, and a distributed fusion structure of global information feedback is designed to overcome the shortcomings that all the information in the simple convex combination fusion algorithm and the covariance intersection algorithm are considered together. The simulation results show that the proposed algorithm has less mean square error than the simple convex combination fusion method and covariance cross-fusion method.