一种约化序贯最小二乘解法

A Sequential Least Square Estimation with Eliminating Local Parameters

平方根最小二乘法是解决传统最小二乘估计中法方程奇异的有效方法之一，但是，当观测方程具有大量局部参量时，直接使用该方法并不可取。基于等效观测方程思想，本文提出了一种能够约化局部参数的序贯最小二乘算法。

The square root information filtering is one of the useful method for solving the singularity of normal matrix for the least square estimation. However, this method can＇t be used directly when there are many local parameters in observation equation. Based on the idea of equivalent equation, this paper presents a sequential least square estimation which can eliminate the local parameter.