Optimized Network for Predicting Total Iron Content and Alkalinity of Sinter with Input and Middle Layers Nodes Reducing

Abstract

Chemical indicators have always been important parameters to evaluate the quality of finished sinter. At present, however, most sintering plants have no good control over sinter composition. Due to the too long regulation cycle, the analysis results of data cannot guide the ingredients timely. Thus, the advanced prediction of chemical composition of finished product sinter is needed, in order to timely optimize and improve the grade of sinter. The BP neural network is used to establish the prediction model of the chemical composition of sinter. With the strong adaptive and self-learning ability, it can adjust the connection weights of the network in time by referring to the online detection data, and keep the dynamic change of the sinter production system in order to improve the accuracy of the prediction. The simulation results show that good efforts are realized on total iron content and alkalinity prediction.