线性系统的几种辨识算法

Algorithms of Linear System Identification

本文提出了线性系统的几种辨识算法,在连续系统辨识中,改进的辨识算法,其精度与速度均高于传统算法。在最小二乘辨识算法中,不用矩阵求逆运算,使辩识速度有很大提高。

In this paper, an algorithm is presented to evaluate the frequency response from the known impulse or step response data using a very efficient numerical differentiation method, compared with the conventional algorithm, it improved the speed and accuracy in continuous system identification. An improved batch LS identification algorithm is presented which avoids matrix inversion successfully and speeds up the identification process. It is especially suitable while the order of the system is unknown. In this paper an instruction of a software module based on the above algorithms is also presented.