Application of artificial neural network model to predict reduction degradation index of iron oxide pellets

Abstract

The reduction degradation index (RDI) is an important metallurgical property of iron ore pellets used for the production of RDI from shaft furnace or for use in blast furnaces. In order to develop a control strategy, a neural network model has been developed to predict the RDI of pellets from 13 input variables, namely feedrate of green pellets, bed height, burn through temperature, firing temperature, specific corex gas consumption, bentonite, moisture and carbon content in green pellets and Al₂O₃, SiO₂, CaO, MgO and FeO in fired pellets. The RDI of pellets was more sensitive to variation in MgO, CaO, bentonite and green pellet carbon content. The predicted results were in good agreement with the actual data.

Keywords:

• NEURAL NETWORK
• REDUCTION DEGRADATION INDEX
• PELLETISATION
• INDURATION
• GREEN PELLET
• BURN THROUGH TEMPERATURE
• BENTONITE
• PELLET QUALITY