A Desulfuration Prediction Model Research Based On RBFNN

Sulfur is the primary nocuous element in iron. In order to achieve the standard request, desulfuration is the most important task in iron production. Improving the desulfuration standard and decreasing the desulfuration agent consumption are primary targets in desulfuration of steel production. Under high temperature atmosphere, magnesium which has been widely used in iron pre-desulfuration has perfect appetency with sulfur. Single-spouted magnesium technology, which has a high efficiency of desulfuration, the rapid reaction rate, the less consumption, a short disposal time, and the low temperature descension of molten iron after management, is widely used in iron pre-desulfuration. The effect of desulfuration in iron pretreatment is related to the steel quality in production. Magnesium consumption is vital in the desulfuration process. At present, the prediction model which could confirm the quantity of magnesium was reported rarely.The research based on the background of the molten iron desulfuration project of New FuShun Iron Co. Ltd. The MAMDANI fuzzy neural network (FNN) is created based on the productive swatchs and experimental regulation. The FNN can either guarantee the trend of the regulation while general consequence, or have the perfect close precision while history recollection. The rational data were verified. The results indicated that the data whose error was in 25% up to 85.7%, and the error less than 10% up to 78.6%. So, it could be used to direct the production.