多变量CARMA模型的结构辨识
Structure Identification of Multivariable Carma Models
-
摘要: 本文提出了多变量受控的自回归滑动平均(CARMA)模型结构辨识的新方法.根据模型参 数的递推增广最小二乘法(RELS)估计,给出了确定模型的阶、子阶和时滞的F检验判决器, 且可得到节省参数模型.本方法推广和改进了Bokor和Kevizky的结构辨识方法, 数值模拟例子证明了所提出方法的有效性.Abstract: In this paper, a new approach of structure identification for multivariable CARLA models, is presented. Based on the recursive extended least squares (RELS) estimation of the model parameters, the F-test detectors for obtaining model order, suborder and time delay are given, and the parsimonious parameter models are also given. The structure identification approach of Bokor and Keviczky is extended and improved. A numerical simulation example is given to illustrate the usefulness of the proposed approach.
计量
- 文章访问数: 1937
- HTML全文浏览量: 90
- PDF下载量: 1107
- 被引次数: 0