Controle neuro-fuzzy aplicado para a redução de vibrações em uma viga compósita

Abstract

This study focused on investigating the active control of vibrations in composite material structures, employing the Neuro-Fuzzy intelligent control strategy. The computational implementation was entirely performed on the MATLAB® platform and using the Simulink tool. The control system consists of three main components: the mathematical model of the beam identified using the ERA/OKID method, a state estimator, and the controllers. Based on the controllability and observability analysis, the system was reduced to the first two modes. After the numerical study of the control system, the methodology was experimentally validated. This experimental stage was conducted in two parts. Initially, a pole placement controller based on the Ackermann method was developed. The input and output data from the pole placement controllers were used in the second stage to train the Neuro-Fuzzy controller. The data used in this training consisted of displacements and modal velocities as input variables and the corresponding electrical currents as system outputs. Thus, both pole placement controllers and Neuro-Fuzzy controllers were analyzed. The experimental results demonstrated the efficiency of the proposed methodology, given the reduction in the vibration level in the response of the analyzed system. The Neuro-Fuzzy controller proved to be particularly advantageous due to its ability to combine adaptive learning of artificial neural networks with Fuzzy Logic, resulting in a robust and computationally efficient control system for active vibration control applications in flexible structures.