Research On Lf Steel Temperature Prediction Base On Hybrid Model Using Xgboost

The refining process of refining furnace(LF)is the most important process in steelmaking,it has been widely used in steelmaking enterprises in China.The most important factor in LF refining process is temperature control.How to accurately control the temperature of molten steel has always been the focus of research.Accurate control of molten steel temperature in LF refining process is helpful to improve the quality of steel,reduce the cost of steelmaking,select the best control strategy and reduce the risk of personnel operation.Therefore,accurate prediction of molten steel temperature is of great significance for LF refining.Firstly,the LF refining process and the factors affecting the temperature of molten steel are well studied.In view of the shortcomings of the expert system prediction model in practical application of steel plant,and the advantages of machine learning algorithm,a hybrid prediction model based on XGBoost is proposed.It realizes parallel prediction of intelligent model and expert system prediction model.This hybrid prediction model not only improves the estimation accuracy of molten steel temperature,but also enhances the adaptability of the algorithm.Finally,this model is trained and tested with the actual production data.The experimental results demonstrate that the proposed model has high estimation accuracy and good applicability.Due to the complexity of XGBoost algorithm,there are many parameter optimization problems in the learning process,among which different combinations of parameters may cause great differences in the accuracy of the model.The traditional random search method and grid search method are very inefficient,which makes the algorithm uncertain and stochastic.In this paper,Bayesian Optimization Algorithm(BOA)is used to optimize the parameters of the intelligent model,and a hybrid prediction model of BOA-XGBoost and expert system is constructed.The experimental results show that the effect of BOA parameter optimization is better than that of traditional methods.The precision of the optimized hybrid model has been improved obviously,which is more suitable to the production needs of enterprise.Finally,the optimized hybrid prediction model is applied to the LF refining system,and a temperature prediction function module is designed,which is tested with actual production data.The test results show that the temperature prediction function module has high accuracy and stability.It is not only suitable for molten steel temperature prediction in LF refining process,but also can provides a reliable reference for accurate temperature control of LF refining process.