Modelling and prediction for steel billet temperature of heating furnace

Since heating furnace is a multivariable non-linear system with large inertia, net lag and crossed coupling, it is difficult to estimate the steel billet temperature in the industry. This paper establishes billet predictive model between billet temperature variable and heating process variable with improved extreme learning machine (ELM) method. At last, it reckons model parameter based on actual data in practice. Analysing check and error indicate that this model can forecast billet steel exit temperature before ten to 25 minutes, and the forecasting error can satisfy industry application accuracy demands.