Learning Wavelet Control Using Constructive Wavelet Networks

Abstract

In this chapter we explore another important direction: combination of ILC and other predominant intelligent control methods characterized by black-box based approximations. In preceding two chapters we focused on systems with parametric uncertainties. In practice many systems are also characterized by non-parametric (lumped) nonlinear uncertainties. In such circumstance, black-box methods may be the best choice. The area of black-box methods is quite diverse, and covers topics from mathematical approximation theory, estimation theory and non-parametric regression, to algorithms and currently widely discussed concepts like neural network, fuzzy and wavelet models [117].