Variable neural adaptive robust controllers for uncertain systems

摘要

A class of variable neural adaptive robust controllers is proposed. Essential components of the proposed controllers are raised-cosine radial basis function (RCRBF) neural networks that can vary their structures dynamically by adding or removing RBFs online. The compact support of the RCRBFs alleviate the problem of determining the parameters of the RBFs and can, together with a simple adding and removing algorithm, significantly reduce the computational effort in the training process of the network. The stability of the overall closed-loop system is analyzed using the Lypaunov type of arguments. It is guaranteed that the tracking error of the closed-loop systems driven by the proposed controller is uniformly ultimately bounded.