Prediction And Implementation Of The End Of The Support Vector Machine-based Converter

The Basic Oxygen Furnace (BOF) steelmaking end-point's automatic control is a key technique in the steelmaking, its aim is to control the carbon content and temperature of the furnace steel to meet the requirements. It was found that the furnace mouth flame spectral data implied the information of carbon content and temperature of the furnace steel. For this reason, a forecast model of the BOF endpoint was designed on the basis of the furnace mouth flame spectral data. This model uses support vector machines (SVM) which has strong learning ability and needs only less number of samples as its learning algorithm, and uses the characteristic quantities which could characterize the furnace mouth flame spectral data to improve the efficiency of model training. Training,testing and forecasting of this model are realized using MATLAB simulation software. To fillful the requirements of real-time engineering applications, the BOF online endpoint prediction system based on virtual instrument technology was designed of which the core is the BOF steelmaking end-point prediction model based on SVM. The simulation and online experiment demonstrate that the BOF steelmaking end-point prediction model based on SVM could evidently increase the end shooting rate, and the BOF online endpoint prediction system has good predictive ability and generalization ability.