Self-Tuning Multi Step Prediction of Strength Index in an Iron-Ore Sintering Process

Sintering is a complex metallurgical process by which iron ore is converted to sinter before it is fed to the blast furnace. An important property of the sinter is its strength and present practice requires the plant operator to make adjustments to the input variables as he considers necessary, based on off-line strength tests and on his accumulated experience and empirical understanding of the process. Although the operator may be highly skilled, the complexity of the process and the long time constants involved, give rise to relatively crude control of sinter strength. In recent years, significant advances have been made in the area of self-tuning prediction and regulation of stochastic processes and the methods have been applied to a number of industrial problems. In manually controlled processes better results can be achieved by providing the operator with advance information on predicted future performance of the process, which is calculated on the basis of past and present performance together with proposed future input strategy. The work reported here represents the first known attempt to develop a self-tuning multi-step predictor for sinter strength.