Predicting the End Temperature of Molten Steel Using CBR

Abstract

Steel is a kind of important material. The accurate control about the end temperature of molten steel has significant impact on the quality of steel material. Case Based Reasoning (CBR) is used to predict the end temperature of molten steel in Argon Oxygen Decarburization (AOD). Grey Relational Degree (GRD) with different weights of attributes is adopted to calculate the similarity between cases. Analytic Hierarchy Process (AHP) is taken to determine the weights of attributes. Multiple Linear Regression (MLR) is applied to compute the relative weight of two different attributes for AHP. Two methods, CBR using AHP with Equal Weights (CBR_AHP_EW) and CBR using AHP with Different Weights (CBR_AHP_DW), are employed to for a comparison. The results show that CBR_AHP_DW is effective in predicting the end temperature of molten steel in AOD and CBR_AHP_DW outperforms CBR_AHP_EW.