A novel method for endpoint temperature prediction in RH

Abstract

ABSTRACTBased on Cloud Model, a novel method was proposed to predict the endpoint temperature in Ruhrstahl Heraeus (RH) for Interstitial-free (IF) steel production, considering the starting temperature, scrap and refining cycle. 300 sets of RH production data was collected, mined and reasoned by Cloud Model. The prediction results of the Cloud Model are compared with BP neural network methods. The final results show that in the error scope from −10 to 10°C, Cloud Model acquired the 93.32% hit rate, BP neural network acquired the 89.33% hit rate. Compared with the BP neural network, the Cloud Model has higher accuracy and stronger generalisation ability.