数据融合系统中,传统系统误差估计算法在某些特定物理场景中表现出显著的病态性,估计结果不收敛。为了解决这个问题,该文首先深入分析产生算法病态性的物理场景并将其分为传感器密集型和目标密集型2类,然后将系统误差分解成相对方位系统误差和剩余系统误差2部分分别估计。理论分析和仿真实验表明:该方法在估计相对方位系统误差部分不存在算法病态性;在传感器密集型物理场景中,该方法在估计剩余系统误差部分对病态性进行有效控制;在某些物理场景中,该方法的算法条件数可缩减为传统方法的1%,方位系统误差估计样本方差可缩减为传统方法的1/5。 In the data fusion system, the traditional system error estimation algorithm shows significant morbidity in some specific physical scenarios, and the estimation results do not converge. In order to solve this problem, the paper first deeply analyzes the ill-formed physical scene of the algorithm and divides it into sensor-intensive and target-intensive class 2, and then decomposes the system error into two parts: relative azimuth error and residual error . The theoretical analysis and simulation results show that this method has no ill-posedness in the estimation of systematic errors in relative azimuths. In sensor-intensive physics scenarios, this method effectively controls the ill-posedness in estimating the residual systematic errors. In some physical scenarios , The number of algorithm conditions for this method can be reduced to 1% of the traditional method, and the variance of the azimuth systematic error estimation sample can be reduced to 1/5 of the traditional method.