A TWO-LEVEL SELF-TUNING PREDICTION METHOD AND ITS APPLICATION TO Si-CONTENT PREDICTION OF MOLTEN IRON IN BLAST FURNACE

A two-level self-tuning prediction method is proposed for a blast furnace,a complex MISO time-varying system.Combining the recursive identification of the faster time-varying part with the iterative correction of the slower part,this method solves the problem of too big disparities in time constants andtime-varying speeds of submodels.Practice shows its effectiveness for predicting Sicontent of molten ironin a blast furnace.