Prediction of Iron Ore Sintering Characters on the Basis of Regression Analysis and Artificial Neural Network
Abstract
Iron ore sintering is a complex and hysteresis process, so the prediction of the iron ore sintering character is very necessary. By using stepwise regression analysis, this paper confirms the main factors that influence moisture content, fuel ratio, sintering speed and the sintering drum strength. By using BP artificial neural network, we construct the prediction model against the four characters mentioned afore, and prediction accuracy are 96.67%, 93.33%, 86.67% and 93.33% respectively. The model result used in sintering pot test, the experimental results show that the mix moisture and fuel ratio are optimized, and the sinter drum strength is improved.
1 Preamble
The iron and steel are the most vastly used structural materials and the biggest output functional materials, which amounts to 626,700,000 tons in 2010 in China, 42% of the global output. Along with iron and steel industry's