Abstract
This paper presents the results of an investigation into the use of GMDH neural networks for system modelling and prediction. A number of Adalines with nonlinear preprocessors were trained to yield a GMDH neural network. The training was carried out by applying the Widrow-Hoff learning rule. The structure of the network, i.e. the number of layers and the number of Adalines in each layer, was determined during training. The results obtained have shown that this type of network can successfully be employed for various system modelling and prediction tasks