Choice of models for on-line identification of MIMO stochastic systems

Abstract

Various model forms for the on-line identification of multi-input multi-output (MIMO) stochastic dynamic systems are reviewed. These models are the transfer function matrix, input-output difference equation, state-space representation, and decomposition form. The number of parameters to be identified and the structural parameters required to characterize each model are tested and compared. Results show that the decomposition form is very powerful for solving structure and parameter estimation problems of MIMO systems.