{Desenvolvimento de um Controlador Preditivo Generalizado Otimizado com Algoritmo Genético para Rastreamento de Sinais Senoidais com Nível CC

Abstract

Model Predictive Control (MPC) is one of the most advanced control techniques that can be used in different industrial applications. One of the techniques developed under this definition is called Generalized Predictive Control (GPC), whose objective is to obtain a set of future control laws that are capable of reducing the error between the predicted outputs and the future references of the plant. However, the traditional GPC technique is not capable of being used to trace and/or reject sinusoidal references, which have great importance in practical applications. Furthermore, some applications require that the sinusoidal reference also contain a DC level, or step disturbances to be rejected. Thus, this work is aimed at the development and experimental implementation of a GPC algorithm that is capable of tracking references composed from a combination of step and sinusoidal signals as well as the usage of genetic algorithms in order to determine the optimized implementation parameters of a GPC controller associated with a plant under the influence of white gaussian noise. Simulation and experimental results confirm that the technique proposed here is capable of adequately tracking these types of references, and also has the ability to reject sinusoidal and step disturbances.