Research On The Fuzzy Modeling Of Blast Furnace Gas Flow Based On The Fusion Of Mechanism And Data

Blast furnace(BF)ironmaking process is the upstream process of metallurgical industry.BF is the main equipment of high energy consumption and high environment pollution.The long-term stable and smooth operation state of BF is the key to reduce energy consumption and environmental pollution,and it is the key to ensure the production and quality of molten iron.Therefor,it is related to the industry competitiveness and economic benefits of steel companies.The data mining,feature extraction and modeling prediction of measurable data in the BF process are important means to avoid BF abnormal states and maintain stable operation state of BF.The production environment of BF is very odious,some important field detection information can't be obtained,makes it difficult to analyze the furnace conditions by detecting the internal information of BF.BF gas flow participates reduction reaction and heat transfer during the BF process,it is closely related to the operation states and internal temperature field distribution of BF.And BF gas flow carries a lot of BF states information.With good timeliness,it takes only a few minutes flowing from the bottom to the top in furnace.Therefor,it has important practical value to use the information of BF gas flow in the process of BF states prediction.According to the mechanism and expert experience of BF,this paper analyses the change signs of some related indexes of BF gas flow before the occurrence of abnormal BF conditions which are caused by the abnormal distribution of BF gas flow.Based on the gray correlation algorithm,the degrees of gray correlation in the indexes are computed and then the redundant indexes are eliminated.Furthermore,the maximum information coefficient method is used to add the time-delay variables.In recent years,T-S fuzzy model has achieved great success in the complex industrial systems modeling,forecasting and model-based control.For the data feature extraction of T-S antecedent part identification based on one cluster prototype is incomplete,in this paper,the T-S fuzzy model of weighting rule firing level is proposed based on the hyper-spherical-shaped cluster prototype and thehyper-plane-shaped cluster prototype.This modeling method can extract more important data features,thus to improve the modeling quality and the prediction accuracy of T-S fuzzy model.Then the T-S fuzzy model of weighting rule firing level proposed in this paper is used to forecast the important indexes of BF gas flow,such as bosh gas volume,hot air pressure and northeast temperature of furnace top.Algorithmic program is wrote by the MATLAB simulation software.The simulation experiments on a public test dataset and the BF historical dataset show that the model proposed in this paper has a strong generalization performance.Compared with the T-S fuzzy models based on one cluster prototype,the improved T-S fuzzy model has a higher prediction accuracy.Finally,based on the C# programming language and SQL Server database technology,the “blast furnace indexes prediction and furnace condition decision system” is finished.The interface of this system shows the real-time data and prediction results of abundant BF indexes,it displays the current BF operation information intuitively.This system makes it convenient for the operators in the BF field to observe and analyse the operating conditions of BF.In the field of 2# furnace,Liuzhou steel company,this system had been tested and runs for a long time,the prediction effect is well.