Parameters adjustment for VOD endpoint carbon content and endpoint temperature prediction model

Abstract

In Vacuum Oxygen Decarburization(VOD) steel refining process, the endpoint carbon content and endpoint temperature are criteria for smelting products. A VOD model is often needed to predict the endpoint data. During the modeling of VOD, some parameters are difficult to chose, thus affects the model prediction accuracy. Based on the VOD mathematical model, the process is analyzed to chose the main factors inflecting the prediction. Using RBF neural network, the correlation parameters is adjusted to improve the model forecast accuracy. The simulation result shows the prediction accuracy is better than it does before.