ANNNAC – extension of adaptive backstepping algorithm with artificial neural networks

Abstract

The adaptive backstepping algorithm is a well-known scheme for the design of nonlinear adaptive controllers. The two main drawbacks associated with this algorithm are that the nonlinear system must be linearly parameterised in the unknown or uncertain parameters and that the nonlinear functions must be exactly known. To avoid these problems, an extension of the backstepping algorithm with a specific type of artificial neural networks (ANN) called radial basis function networks (RBF), is proposed. This extension leads to a new control scheme; namely artificial neural network nonlinear adaptive control (ANNNAC). To further clarify the approach, a simple example is studied and the simulation results demonstrate clearly the power of this extension.