Controle de sistemas não-lineares com o uso de redes neurais e garantias de robustez

Abstract

The development and improvement of control techniques for nonlinear systems is an inconclusive problem extensively addressed in the literature. Seeking the best way to assure that a given output of a nonlinear system follows a reference, several methodologies are proposed in the literature, ranging from linearization around a reference or trajectory to techniques aimed at dealing with the lack of linearity. With the advancement of technology, the integration of control techniques with neural network concepts has stood out in several studies, significantly expanding the possibilities in controller modeling. Neural networks allow simulating the behavior of controllers, eliminating repetitive calculations and significantly reducing operational costs. Moreover, they generate robust and adaptable control signals. In this context, the present work proposes the analysis and simulation of neural network-based controllers.