Controle neuro-fuzzy aplicado à estruturas mecânicas utilizando atuadores eletromagnéticos

Abstract

The present work aims to apply active vibration control in a two-degree-of-freedom structure. When vibration becomes a problem causing failures, vibration control becomes an ally to correct them. One of the methods used is active vibration control. In this work, Fuzzy and Neuro-Fuzzy Theories were used to design the controllers and compare the results. In the first stage, the work consisted of performing numerical simulations in the MATLAB® Software, in which the mass-spring-damper system model was implemented with two degrees of freedom. Three cases were considered for analysis: The LQR controller with the Inverse Model of the Electromagnetic Actuator, the Neuro-Fuzzy Controller with the Inverse Model of the Electromagnetic Actuator and finally, the Neuro-Fuzzy Controller without the Inverse Model of the Electromagnetic Actuator. The first case showed better accommodation time; in the second case, the Control Mode 2 obtained a better attenuation and the last case, even removing the Inverse Model, also obtained good results in the attenuation of vibrations. With the numerical results, it is possible to conclude the validity of the proposed methodologies for the active control of vibrations. In order to validate the proposed methodology after the numerical simulations, a second step was carried out, which consisted of experimentally validating the controllers designed in a bench with two degrees of freedom. The experimental results were satisfactory in the active vibration control, demonstrating the validity of the designed controllers