Desenvolvimento de controle particionado com inferência fuzzy aplicado a sistemas não lineares

Abstract

This work presents a non-linear system control technique. It is based on the combination of the control theory for linearized systems and the application of an fuzzy inference to act as a supervisory controller, which analyzes the analyzed situation and indicates the best combination of control for each situation. To develop this application, different linear controllers were designed for different regions of the phase plan, configuring a keyed control system. With the projected controllers, the fuzzy inference is used to analyze the system state variables. With this, it is possible to perform an interpolation of linear controllers to be applied in a non-linear system, in order to smooth the control close to the limits between regions. This technique is implemented in a system of a simple pendulum with an operating point around or (in this work called the inverted pendulum), as it has non-linear dynamics and is largely studied in the control literature.