Comparative Study of Three Output-Error Multi-Input Multi-Output Identification Methods
Multi-input multi-output (MIMO) systems have been a major concern for decades. However, due to the intrinsic complexity raised by the process interactions and optimization issues, MIMO approaches have not been developed as extensively as the single-input single-output ones. Recently, nevertheless, several algorithms have been proposed to address this problem, most of them based on recursive algorithms and many dependent on the assumption that the transfer function denominator polynomials are the same for all subsystems. In this article, an iterative least-squares-based algorithm, a pseudolinear regression and a Gauss-Newton optimization-based algorithm are proposed to provide a continuous-time output-error multi-input single-output model by means of iterative strategies. The numerical simulations indicate the iterative least-squares-based and the pseudo-linear regression algorithms have similar performances and generate more accurate and precise estimates than the Gauss-Newton one, which presented averages and standard deviations of the parameters ranging from twice as large to one order of magnitude higher than those of the other two algorithms.