Hierarchical estimation algorithms for multivariable systems using measurement information

Abstract

With the development of industry information technology, many modelling methods have been focusing on the estimation problems of multivariable systems, especially for the multivariable systems with output error autoregressive noises, from input–output measurement information. Since such a system includes both a parameter vector and a parameter matrix, the conventional methods cannot be applied to parameter estimation and modelling. In order to solve this difficulty, a hierarchical least squares based iterative identification algorithm and a hierarchical generalized least squares identification algorithm are proposed. The basic idea is to decompose the system into two fictitious subsystems, to estimate the parameters of each subsystem, and to coordinate the associated items between the two subsystems. The simulation results indicate that the proposed algorithm is effective.