Research And Application Of Optimization Model For Alloy Feeding System Of LF Furnace

LF(Ladle funace)is the key equipment for secondary refining in steel smelting and plays an important role in improving steel quality and production efficiency,As the core step of LF furnace refining technology,alloying opeeration not only needs to precisely control the content of alloying elements in the molten steel to reach the standard,but also needs to control the content of impurity elements within the standard range.According to the minimum cost of adding materials,more accurate control of the composition of molten steel content,help to improve the production efficiency of iron and steel enterprises,reduce production costs.This paper takes a LF furnace feeding system of a steel enterprise as the research background,combines the actual smelting data and production process,analyztes the key points and difficulties of the alloying operation problem in the LF refining process,and designs a LF furnace alloy feeding model.First,this paper uses the Extreme Learning NMachine(ELM)algorithm to establish the prediction model of alloy element yield,and compares the model)with the model established by BP algorithm.The results show that the prediction results of ELM algorithm are more in line with the expected results and the prediction effect is better.However,with the decrease of sample data provided to the model,the prediction effect of ELM model fluctuates greatly and the stability of the model decreases.In order to optimize the model,this paper improved the crossover operation and mutation operation of the algorithm on the basis of the adaptive genetic algorithm(AGA).The improved AAGA algorithm can dynamically adjust the crossover probability and mutation probability of the algorithm according to the difference degree of the better individuals in the population evolution to avoid the premature of the algorithm.In order to verify the improvement effect of the algorithm,this paper firstly uses the AGA algorithm to optimize the BP model and the ELM model respectively,and then uses the AAGA algorithm to optimize the ELM model and compare the experimental results of multiple models.The experimental results show that the AAGA-ELM element yield prediction model has the highest accuracy and the best model stability.Then,using the linear programming method and AAGA-ELM model,the LF furnace alloy feeding model based on minimum cost is established.This model is used to precisely control the content of alloy elements in the molten steel,and at the same time to control the content of impurity elements not exceeding the standard,and to minimize the feeding costFinally,this paper applies the LF furnace alloy feeding model to the actual production of the enterprise and embeds the alloy feeding module in the existing LF secondary control system,which can realize the alloying operation in the actual production.