By Syed Saad Azhar Ali, Hussain N. Al-Duwaish
.
Keywords: Wiener model, multivariable, Radial Basis Function Neural Network.
A new method is introduced for the identification of the nonlinear multi input multi output (MIMO) Wiener Model, comprising of linear dynamics in cascade with static nonlinearities. The static nonlinearities are modeled by Radial Basis Function Neural Networks (RBFNN) and the linear part is modeled by MIMO autoregressive moving average (ARMA) model. The new algorithm makes use of the well known mapping ability of RBFNN. The learning algorithm is an extension of SISO identification scheme presented in [1] . The proposed algorithm estimates the weights of the RBFNN and the coefficients of ARMA model based on least mean squares (LMS) principle simultaneously