Determination Of End - Point Carbon Classification Of Converter Steelmaking Based On Spectral Analysis And Support Vector Machine Algorithm

The endpoint control of carbon content is a key process in the production of BOF steelmaking, which is directly related to the quality of steel, steelmaking costs and exhaust emissions. However, in the smelting process, due to the instability of the raw material is added and complex chemical reactions in the process of blowing, the endpoint accurate control of carbon and temperature has been a worldwide problem be solved for many years in metallurgical industry. Therefore, the study of a high precision detection method of online and real-time is particularly urgent. Aiming at this problem, a real-time classification detection method of carbon content of steel is studied.This paper analyzes the research of BOF endpoint control technology at this stage, describes some current detection methods, at the same time, summarizes their pros and cons. This method is based on the spectroscopy of flame at furnace mouth and the principle of support vector machine. Firstly, this paper analyzes the spectral of flame and extracts characteristic variables. Secondly, based on the SVM classification algorithm, characteristic variables are trained, and classification model can be obtained. Finally, this paper designs a real-time detection algorithm of steelmaking process, the exact value of carbon is detected by existing fitting curves, and based on the Lab VIEW platform, develop a real-time detection system at the scene.This method is proved to be high-precision, real-time detection and strong anti-jamming ability, etc. Entirely suitable for site environment, and meet the practical requirements of steel.