多传感器误差配准研究

多传感器组网信息融合时，需要对组网中各传感器系统误差进行估计和补偿，以消除各传感器系统误差对融合性能的影响。研究了雷达组网系统误差配准模型，并对最小二乘算法 (LS)、广义最小二乘算法(GLS)、递推最小二乘算法(RLS)、修正 EX 算法等误差配准算法进行对比分析，同时给出了扩维配准模型用于解决多传感器组网配准问题，针对实际工程应用中算法收敛情况难以判断的问题，提出了一种误差配准收敛性判定的方法。根据仿真结果对误差配准算法性能进行分析，同时验证了扩维配准和收敛性判定方法的有效性，支撑配准系统应用。

Registration for Multi-Sensor Networking System

Registration is a prerequisite process for the multi-sensor networking system to estimate and correct systematic errors accurately. The classical registration algorithm such as least square algorithm (LS), generalized least square algorithm (GLS), recursive least square algorithm (RLS) and corrected exact method algorithm (CEX) are compared. An extended registration model is proposed to solve the registration problem for the multi-sensor networking system. The convergence of algorithm in real-time registration is difficult to judged, so this work provides a judgment method. The result shows the performance of such registration algorithms and confirmed the validity of extended registration model and convergence judgment method, the guidance is given in the practical choice of registration algorithms.