Identification of nonlinear systems using Haar wavelet networks

Using Haar wavelet functions, we design a wavelet network for adaptive identification and control of the nonlinear dynamical systems. For approximating nonlinear functions and system identification applications, wavelet method is a good alternative to a neural network. The orthonormal properties of Haar wavelet functions result in a simple training phase. When comparing with the neural network approaches, we have shown that this wavelet approach offers significant improvement in terms of speed, reliability, and accuracy.