Computation And Optimization Of Converter Steelmaking Feed Quantity Considering The End Point Carbon Content And Temperature

Achieving the end point target of converter steelmaking is one of the most important processes in steel production.Whether the target of the end point carbon content and temperature(EPCT)can be accurately achieved will directly affect the quality of steel.The EPCT is mainly determined by the quantity of feed,so the calculation of the feed quantity is very important.At present,manual experience method and static model method are widely used in converter steelmaking process.The converter steelmaking process is characterized by many influencing factors,short smelting process and large quantity of information.Using machine learning and intelligent optimization methods to study the calculation of the feed quantity can avoid the problems of artificial experience participation and insufficient accuracy of static model.The main research contents of this dissertation are as follows:Aimed at the problem of computation of the initial feeding quantity in converter steelmaking,this dissertation first preprocesses the actual data,and selects the appropriate sliding window length through random search method.Further,this dissertation selects samples through the sliding window,and establishes a prediction model based on the XGBoost method,which is compared with other methods by experiments.The results show that the model has higher prediction accuracy and obvious advantages in training speed.Finally,based on the prediction model,a random correction scheme based on positive and negative correlation is designed to correct the initial feeding quantity.The results show that this correction scheme has more advantages than the complete random correction,and a better initial feed quantity can be obtained through this scheme.In order to construct the optimization model of supplementary feeding quantity,the regression between supplementary feeding quantity and EPCT is studied.First,an improved ensemble learning method is proposed based on the Genetic Algorithm based Selective Ensemble(GASEN).Then,based on actual data preprocessing,a regression model between the supplementary feeding quantity and the EPCT is constructed by using this method,and a comparative experiment is carried out with GASEN and other methods.The results show that the fitting effect of the regression model is good,which lays a foundation for the construction of the optimization model of supplementary feeding quantity.Aimed at the problem of optimization of supplementary feeding quantity in converter steelmaking,this dissertation first constructs the optimization model of supplementary feeding quantity according to the regression model of supplementary feeding quantity and EPCT.Then,a Moving-Non-dominated Sorting Genetic Algorithms-Ⅱ(M-NSGA-Ⅱ)is designed according to the requirements of computation time in converter steelmaking process.Finally,a comparative experiment with other algorithms is carried out.The results show that the optimization model based on this algorithm can obtain a better supplementary feeding quantity in a shorter computation time.