Técnicas de otimização H∞ utilizando LMI.

Abstract

Many industrial processes are multivariable, and subject to strong control especifications. In these processes, the representation of systems by mathematical models is not perfect due to the inherent uncertainties in the modeling or identification process, caused by simplifications or unconsidered dynamics. With the application of robust control techniques, it is possible to guarantee closed-loop stability and system performance specifications even under the influence of model uncertainty and external disturbances. The main technique used in the design of robust controllers is the minimization of the H∞ norm, which can be formulated as a problem of convex optimization using linear matrix inequalities. In this work, some techniques are studied for the design of optimal and robust controllers, using the approach of linear matrix inequalities for solving. The application of H∞ optimization method for robust centralized control project is treated, considering systems represented by a set of non-parametric uncertainties. A robust decentralized control design methodology using an iterative procedure is proposed. To mitigate the attenuation of external disturbances, a feedforward control design methodology is proposed. For validation of the proposed methodologies, simulations and experiments are applied in the control project for a temperature multivariable system.