Controle com adaptação em modo dual utilizando uma rede RBF

Abstract

This work presents a control strategy using a network with radial basis function (RBF network) with adaptation in dual mode. The objective of the strategy is to use the approximate capacity of the RBF network to control nonlinear systems with unknown parameters or with uncertainties. The proposed control uses the structure of Model Reference Adaptive Control (MRAC) and a RBF network whose parameters are adjusted in real time in dual mode, which will allow a fast transient and a smooth control signal in steady state. The dual mode adaptive method of RBF network parameters uses the hyperbolic tangent function, which during the transient provides a similar behavior to variable structure control, and integral laws of MRAC that are regulated by a hyperbolic secant function during steady state. A hyperbolic tangent is used instead of signal function what reduces the chattering phenomenon. A hyperbolic secant is used to regulate the integral law, increasing its effects on steady state and reducing on transient time. It is presented a Lyapunov proof for dual mode method and comparisons through simulations.