Neural network and time series identification and prediction
Zeineb Neji, Feriel Beji · 2000
For some classes of nonlinear systems or time series, an operating point dependent NARMA model can be used to present the system. In this paper we attempt to design artificial neural networks that can help in the automatic identification and prediction of such model, for this purpose, we use the Extended Sample Autocorrelation Function (ESACF) as a feature extractor for the network identification and the robust ANN filter for the robust prediction. The network is tested via different noise level in the identification and prediction process to show the accuracy of the connectionist approach and its robust estimation.