Research On Setting Optimization For LF Alloying Addition Based On SVM Algorithm

Ladle Furnace is an important process equipment for refining molten steel.It plays an important role in the alloying control of the LF.However,for a long time,the current optimization model of the set of alloy addition has not achieved good results in practice application.People rely on artificial experience to give the setting value of the alloy addition.This cannot reasonably optimize the alloy feeding scheme.It is harmful to the control of the molten steel narrow composition.This can even affect the quality of the steel.Therefore,this thesis conducts an in-depth research on the optimization model of alloy addition in refining furnaces.Based on the analysis of current optimization methods for alloy addition,an alloying control method is established based on a linear programming model,aiming at the goal of the lowest alloy cost.The model is based on the conservation of materials.Considering that the yield of alloy elements has a great influence on the accuracy of the model,it is difficult to accurately obtain the yield of alloy elements.After analyzing the factors affecting the yield of alloying elements,a prediction model of alloying element yield is established based on the support vector machine optimized by genetic algorithm,Considering the number of the new samples increasing with the molten being smelted,it is also increasing for the range of training setting of elemental yield prediction models.In order to solve the extending of the model prediction time caused by the grow of the training set and improve the prediction accuracy of the model,the HS-SVR incremental learning algorithm is used in this article,which is used to reduce the number of training set samples of the element yield model.The predicted value of the alloy element yield can be calculated by the HS-SVR algorithm.The optimal batching scheme is obtained combined with the linear programming model and the element yield model.The optimal batching model,element yield model,and incremental learning model together form the LF alloy addition optimization setting model.The simulation experiment of LF production data shows that the alloy addition optimization setting model can quickly calculate the alloy setting value according to the composition requirements of the steel type,which is beneficial to reducing the production cost and has a further effect on the precise adjustment of the composition of LF molten steel significance.