Multi-objective Optimization Of Blast Furnace Operation Based On Data And Smelting Mechanism

The energy consumption and carbon dioxide emission of steel industry both account for more than 10% of total energy consumption and carbon dioxide emission all over the country.As the most energy-intensive part of the steel industry,a reasonable optimization control is significant for blast furnace ironmaking process,where energy-saving and emission-reduction is of great importance.In the current blast furnace ironmaking process,the primary control operation is divided into the upper discontinuous control of the furnace top distribution and the bottom continuous control of the pulverized coal injection and blast.The major heat release reacts at the tuyere raceway.The coke entered from the furnace top burns violently with pulverized coal in the oxygen-enriched high temperature atmosphere,and then produces gas flow.Because of the delay influence of the burden,as well as the difficult method of the deviation of the burdening control,the adjustment method mainly focuses on the pulverized coal injection and blast and the oxygen enrichment at the bottom of the furnace.In the process of blast furnace distribution,burden calculation is pre-determined by the mechanisim formula of the material composition,adjustment of the propotion of coke and ore only occurs when there is a steady state deviation of blast furnace.The optimization and adjustment between them is independent for fear of some severe accident such as furnace cooling.It is inevitable that the burdening method causes some extra pollution and loss of the fuel.Therefore,a precise cooperative optimization of the furnace distribution and the bottom parameter operation is of great importance.In order to overcome the modeling difficulties caused by the large time delay and multi-coupling characteristics of blast furnace and the direct observation difficulties caused by the harsh environment of high temperature,high pressure and high dust in blast furnace,a multi-objective optimization modeling framework based on comprehensive data-driven modeling and mechanism model is proposed,considering the overall process of blast furnace ironmaking.Taking the typical intermediate observation parameters of blast furnace(i.e.the contents of Si,P and S etc.)as a bridge,the multi-objective optimization model of blast furnace is modeled step by step.First,the decision variables including blast furnace burden and operation parameters such as lower coal injection and blast are set up by using neural network and PCA method,and then the prediction model of typical intermediate observation parameters is established.Then,according to the elements,material balance and smelting mechanism,the fuel ratio,carbon dioxide emission and cost per ton of iron of blast furnace are described mathematically,and strict constraints are determined.The part with difficult mechanism is described by neural network,and the part with clear mechanism is described by strict formula,taking into account the accuracy and feasibility.Coordinated optimization and adjustment of the upper and lower part of the blast furnace have been realized,with a unified consideration of the upper discontinuous control of the furnace distribution and bottom continuous control of the pulverized coal injection and blast.Considering the complexity,nonlinearity and multi-objective characteristics of the proposed multi-objective optimization model,the nondominated sorting multi-objective genetic algorithm II(NSGA-II)is selected to solve the problem,meanwhile the constraints processing method is optimized.The Pareto frontier of the optimized burden and operation parameters is solved and compared with the actual operation data of blast furnace.The validity of the model and algorithm is preliminarily verified in use,which can provide an effective reference for the decision-making and operation of the blast furnace operator.