Research On Modeling And Optimization Algorithm Of Converter Steelmaking Based On Multi-information

Steelmaking endpoint prediction is currently an important part of the converter steelmaking industry that affects performance indicators,and accurate steelmaking endpoint prediction is of great importance to the entire steelmaking process.In addition to carbon and temperature,there are many chemical elements such as phosphorus and manganese in the composition of steelmaking endpoints that affect steel quality.Therefore,while gradually improving the accuracy of carbon and temperature prediction at the end of steelmaking,the modeling study of phosphorus and manganese content prediction at the end of steelmaking is carried out to promote the comprehensive identification of the end of converter and to guide the operators to improve the accuracy of the judgment of steel production,which is of great significance to further improve the hit rate of the end of converter smelting,improve the quality of steel and reduce the production cost.This thesis adopts a data-driven approach,combined with the population intelligence algorithms that have emerged in recent years,to conduct an in-depth study on predictive modeling and optimization of manganese and phosphorus contents of converter endpoints.(1)Research on swarm intelligence optimization algorithm.The swarm intelligence algorithm has been applied in many fields because of its fast convergence speed and strong optimization ability.The paper first focuses on the research of whale optimization algorithm(WOA)and sparrow search algorithm(SSA).In view of the shortcomings of WOA and SSA in the global search for the optimal solution,this chapter further improves the two algorithms to improve the global convergence ability of the algorithm.For the WOA algorithm,the cauchy mutation operator was used to enhance the global exploration capabilities;the adaptive weight strategy is introduced to improve the local development capabilities of WOA;the differential mutation strategy was introduced to deal with the problem of diversity loss in the whale population at the end of optimization,and an improved whale optimization Algorithm(IWOA)was obtained.For the SSA algorithm,the trigonometric substitution strategy,nonlinear search factor and improved cauchy mutation operator were brought in to address the lack of global optimization of the algorithm,and an enhanced sparrow search algorithm(ESSA)was obtained.Comprehensive evaluation to verify the effectiveness of the method.The proposed improved algorithm will provide algorithm support for the optimization of the prediction model of manganese and phosphorus content at the end of the converter.(2)Prediction of manganese content at the end-point of converter based on IWOALSSVM.Manganese is a very important alloying element in steel.Most of the manganese will be oxidized into the steel slag during the blowing process.Considering the target requirements of the steel grade,it is often necessary to add manganese ferroalloy during tapping or refining.At present,reducing the amount of ferromanganese alloy is achieved by increasing the residual manganese content in the end-point molten steel of the converter and reducing the production cost of steelmaking.In the actual smelting process,there is often a large deviation between the predicted value of the manganese content at the end of the converter and the actual value,resulting in a large investment of ferromanganese alloy,which increases the cost.Therefore,the accurate prediction of the manganese content at the end of the converter is particularly important.Therefore,this chapter used the improved IWOA obtained in part(1)to optimize the least squares support vector machine(LSSVM)model,and obtains a hybrid IWOA-LSSVM end-point manganese content prediction model.The simulation results showed that the hit rate of the end-point manganese content within the error of ±0.01% was 93.3%.(3)Research on end-point phosphorus content prediction in converter based on ESSA-DELM.In the process of converter steelmaking,phosphorus in molten steel is a harmful element for most steel types,which reduces the toughness and plasticity of steel and affects the quality of steel.Therefore,the phosphorus content of molten steel in the converter smelting process of all steel types must be removed to the range specified by the steel type.Therefore,the accurate prediction and control of the end-point phosphorus content of the converter is extremely critical.At first,a deep extreme learning machine(DELM)was used to predict the end-point phosphorus content,but considering the large number of random input weights and biases in the DELM model,the prediction accuracy was reduced.Therefore,the ESSA algorithm in part(1)was introduced to optimize the weights and thresholds,and an ESSA-DELM prediction model was obtained,which was then verified according to the experimental data of the end-point phosphorus content.The experimental results show that the hit rate of end-point phosphorus content based on proposed ESSA-DELM prediction model within the error range of ±0.003%,±0.002%,and±0.001% was 91.67%,83.33%,and 63.55%,respectively.