论文部分内容阅读
在缺少信号或观测的先验分布时,通常使用最小二乘估计参数。而由于实际条件的限制,观测矩阵非满秩,最小二乘估计无效。本文给出了基于奇异值分解的方法,利用Moore-Penrose广义逆,可以得到极小范数唯一的最小二乘解。
In the absence of a priori distribution of signals or observations, least squares estimation parameters are often used. Due to the limitation of practical conditions, the observation matrix is not full rank, and the least squares estimation is invalid. In this paper, a method based on singular value decomposition is given. By using Moore-Penrose generalized inverse, the least square solution of the smallest norm can be obtained.