Stable and Efficient Neural Network Modeling of Discrete Multi-Channel Signals *

This paper presents a neural network based recursive modeling scheme that constructs a nonlinear dynamical model for a given discrete multi-channel signal and the corresponding excitation. Using the so-called Radial-Basis-Function (RBF) neural nets as generic discrete-time nonlinear model structure and the ideas developed in the classical adaptive control theory, we have been able to derive a stable and efficient weight updating algorithm that guarantees the convergence for both the prediction error and the weight error. Elements of the spatial Fourier transform and sampling theory have been employed to provide the guidelines in choosing the parameters associated with the network structure and the RBF neurons.