PREDICTING CHAOTIC SEQUENCES IN A BLAST-FURNACE BY A GENERALIZED RADIAL BASIS FUNCTION NETWORK

Nonlinear forecasting is applied to characterize the complexities in t ime series of temperature fluctuations and pressure fluctuations obser ved in a blast furnace. A generalized radial basis function network an d a Sugihara-May predictor are used as the predictive models. The temp erature sequence may be diagnosed as low-dimensional chaos according t o the scaling property of the prediction error as a function of the pr ediction-time interval. Short-term forecasts about the temperature seq uence are successfully made by the network that has learned the underl ying dynamics embedded in three-dimensional delayed space.