Contribuições ao estudo de sistemas SISO e MIMO com atraso de transporte

• Ano de defesa: 2016
• Autor(a) principal: Correia, Wilkley Bezerra
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Engenharia elétrica; Sistemas de controle
• Link de acesso: http://www.repositorio.ufc.br/handle/riufc/15522

This work presents a contribution to the study of control of dead-time systems. Within this context, it is presented a literature review covering from the Smith predictor until its recent variations. For the state space representation dead-time is normally dealt under the discrete-time perspective, where it is included as state augmentation. In the first case, this work proposes a novel control structure with simple tuning rules, able to deal with single input and output systems that presents multiple delays due to multiple internal paths. However, the design method allows the proposed structure to be readily applied for the single delay case. The study is extended for the state space representation, where it is presented a modification to a predictor recently presented in literature, in order to use it as a predictor-observer within the state space representation. In this case, all the matrices involved in the observer have the same order as those of the plant, i. e., there is no need of the state augmentation to deal with dead-time. Hence, the well known techniques for the optimal and robust control may be applied for multivariable delayed systems. Besides, an alternative way of tuning for the LQG control, based on the T-polynomial approach, is proposed. In this case, the technique may be applied for both the delay free systems and dead-time systems. The latter by applying the modified predictor mentioned earlier in the observer structure. Additionally, it is also proposed a new predictor structure, based on the CARIMA model with T-polynomial, for dead-time systems. This work is full of examples throughout the text which are suitable for both the control strategies studied herein and the proposed ones, applicable for stable, unstable or integrating systems, as well as multivariable systems. Case studies for laboratories plants validate the effectiveness of the discussed methods.