Técnicas de sintonia de controlador difuso aplicado ao controle de processos

Abstract

All real systems exhibits some level of non-linearity in which conventional controllers will not always be able to provide good and accurate performance. As an alternative, fuzzy logic is a well-documented design methodology capable of dealing with such behaviors. It requires, though, adequate tuning, generally achieved via an experienced designer or by heuristics related optimization techniques. In this work, a particle swarm optimization technique is used for the tuning of a fuzzy-PI controller of a water heating system and it's performance to a conventional PI controller. A second process is also studied, of the angular position control of a propeller-attached rod. For this process, three controllers are compared, one obtained via the particle swam method, one via genetic algorithm and another conventional PID. Linearized models of the systems are obtained for the simulations. For each tuning method studied, a set of configurations are tested in order to find the one that better suits each process. The performance of the tested controllers shows high variation according to the selected optimization parameters, highlighting the significance of those and the execution of tests aimed at selecting the most adequate ones for each studied system.