Identificação de modelos de baixa ordem baseados em funções de Laguerre e Kautz com polos otimizados.

• Ano de defesa: 2019
• Autor(a) principal: ANDRADE, Renato Augusto Lira de.
• Orientador(a): BARROS, Péricles Rezende. , CORREIA, Rafael Bezerra.
• Banca de defesa: OLIVEIRA, Alexandre Cunha. , NETO, José Sérgio da Rocha.
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Campina Grande
• Programa de Pós-Graduação: PÓS-GRADUAÇÃO EM ENGENHARIA ELÉTRICA
• Departamento: Centro de Engenharia Elétrica e Informática - CEEI
• País: Brasil
• Palavras-chave em Português: Identificação de Sistemas; Funções de Base Ortonormais; Otimização de Polos; Systems Identification; Orthonormal Basis Functions; Poles Optimization
• Área do conhecimento CNPq: Engenharia Elétrica
• Link de acesso: https://dspace.sti.ufcg.edu.br/handle/riufcg/26799

Low-order models are widely used in the area of ​​systems control. While their intrinsic simplicity is desirable for practical purposes, it is still necessary that the models are accurate representations of the studied system, that is, that there is a minimum level of agreement between the dynamics of the generated model and the real plant. Representations with orthonormal basis functions can be useful in obtaining such models through a finite impulse response model structure, whose use allows to enjoy numerical advantages by being linear in the parameters and estimating the response of the system with dynamics different from that of the system. noise, reducing modeling errors, in addition to allowing the inclusion of prior knowledge about the system in a simple way. However, an adequate choice of poles is crucial in this process. In this dissertation work, we study two situations in which, from input and output data of the system's step response, low-order orthonomal model base functions are generated in order to satisfy the optimality conditions of the estimation error. in relation to the chosen poles. In the first case, it is possible to generate Laguerre models with lag whose poles are calculated in order to satisfy the optimality criteria. In the second, two methods are proposed to generate second-order Kautz models and real parameters, aiming to model systems with dispersed poles that present more complex dynamics induced by zeros. Analytical solutions of the optimization problems that generate the criteria used are also developed.