Application of BP Network and Principal Component Analysis to Forecasting the Silicon Content in Blast Furnace Hot Metal

A novel method for forecasting the silicon content in hot metal is proposed using principal component analysis (PCA) and BP network. PCA can consider the correlations among multiple quality characteristics to obtain uncorrelated principal components. These principal components are then taken as the input parameters of the BP neural network. Then the BP network models are established and trained to map out the functional relationship between the principal components and the silicon content. The application results show that it works well and it is better than BP neural network in efficiency and accuracy, and the hit rate comes up to 86% using the BP neural network and PCA.