Subspace Methods for Dynamic Systems Identification with MOESP

EasyChair Preprint 15351

Abstract

System identification is a pivotal field within control engineering, providing a strategic approach to control challenges when direct modeling of systems proves unfeasible. Using input and output data from the system under analysis, it becomes feasible to construct a mathematical model that captures its dynamics without fully understanding its internal mechanics. Among the various identification methods, the Multivariable Output Error State Space (MOESP) algorithm is distinguished by its straightforwardness, importance, and determinism, enabling the identification of systems represented by state equations. This study explores the steps of this algorithm and its application and validation in dynamic systems. Furthermore, the influence of estimated order, a free algorithm variable, is investigated through graphical and mathematical analysis. We used two examples of dynamic MIMO systems to validate the identification strategy explored in this work. A numerical simulation of a linear system represented the first system. The other constitutes a real nonlinear system. The results presented model evaluation metrics that show the efficacy of the methodology under study.

Keyphrases: MOESP, dynamic systems, state space, subspace methods, system identification

@booklet{EasyChair:15351,
  author    = {Maria Fernanda Silva and Rodrigo Silva and Glauber Leite and Thiago Cordeiro and Ícaro Araújo},
  title     = {Subspace Methods for Dynamic Systems Identification with MOESP},
  howpublished = {EasyChair Preprint 15351},
  year      = {EasyChair, 2024}}