Research On Endpoint Prediction And Control Model For BOF Steelmaking

The iron and steel industry is an important part of the national economy in China which is a traditional steel-making country.China has vigorously promoted the application of new technologies,new equipments and new methods to reduce the cost of steel and improve it's quality,aiming to enhance competitiveness and possess great market share in this field since the 12thFive-YearPlan.However,the technology and equipments fall behind in our small converter currently.The endpoint judgment of converter is still done empirically by workers.Moreover,advanced vice-gun detection technology can not be used as a result of the limits of workshop and economic conditions.Based on the facts above,the development of efficient,practical and economical technique is of great significance.A new method to predict and control the endpoint of converter is put forward in this paper.Firstly,material balance and thermal equilibrium is analyzed to build a static computational model of ingredient-mixing process.This model can realize good management on the ratio of ingredients and improvement of accuracy.Secondly,extreme learning machine(ELM)which can solve non-linear problems easily is used to build a predictive model of the endpoint.Owing to random confirmation of the initial weights and threshold values,the structure of ELM isn't stable.Genetic algorithm(GA)is used to optimize them.Meanwhile,the effects of the different number of hidden layer neurons and different hidden layer activation function to the model prediction accuracy is analyzed.The results of error data and prediction graphics indicate that this improved method is effective.While the static prediction model cannot predict the endpoint accurately in a disturbed environment,so optimal control model based on prediction model is established by particle swarm optimization algorithm(PSO).To enhance the searching ability,the inertia weights of PSO are revised.The effects of the different functions and formulas to the model control accuracy is compared.The simulation results show the method's feasibility.This model can realize the optimal control of the variables which influence the endpoint mostly.Lastly,the proportioning calculation model,prediction model and optimal control model are integrated to a software built by Visual Studio.The prediction and control of steel-making process is realized.