There are two possible general forms of multiple input multiple output (MIMO) regression models, which are either linear with respect to their parameters or non-linear, but in order to estimate their parameters, at a certain stage it could be assumed that they are linear. This is in fact the basic assumption behind the linear approach for parameters estimation. There are two possible representations of a MIMO model, which at a certain level could be fictitiously presented as linear functions of its parameters. One representation is when the parameters are collected in a matrix and hence, the regressors are in a vector. The other possible case is the parameters to be in a vector, but the regressors at a given instant to be placed in a matrix. Both types of representations are considered in the paper. Their advantages and disadvantages are summarized and their applicability within the whole experimental modelling process is also discussed.
1. Efremov, A. Multivariate Time-Varying System Identification at Incomplete Information. Technical University of Sofia, Faculty of Automatics, Ph.D. 2008.
2. Efremov, A. System Identification Based on Stepwise Regression for Dynamic Market Representation. International Conference on Data Mining and Knowledge Engineering, Rome, Italy, 28–30 April 2010, 64, 2010, No. 2, 132-137.
3. Efremov, A. Multivariable System Identification. Monograph, Dar – RH, ISBN 978-954-9489-34-7, 2013.
4. Faraway, J. Practical Regression and ANOVA Using R. http:// cran.r-project.org/doc/contrib/Faraway-PRA.pdf, 2002.
5. Van den Hof, P. M. J. Model Sets and Parameterizations for Identification of Multivariable Equation Error Models. – Automatica, 30 (3), 1994, 433-446.
6. Vuchkov, I. Identification. Sofia, IK Jurapel, 1996.