Modeling Of Burden Distribution Process In Blast Furnace

Iron and steel industry is an important pillar of the national economy,but it has some problems,such as exacerbate contradiction of overcapacity,low level of independent innovation and constraints of resource environment.Hence,in order to improve the energy utilization rate and realize the intelligent smelting,it is urgent to optimize the current production process.Ironmaking is the most significant part of iron and steel industry."High efficiency,high quality,low consumption,longevity and environment protection" is the basic technical policy.The 13th Five-Year Development Plan for Iron and Steel Industry emphasizes the improvement of independent innovation ability,the raising of effective supply level for the iron and steel,development of the intelligent manufacturing andadvancement of the green manufacturing.Hence,realizing the intelligence and the informationization of ironmaking meets the needs of national development.The charging system is one of the four operation systems for blast furnace.Charging system determines the burden distribution.It is a comman method to improve the condition and ensure the smooth operation of the blast furnace.There exist the complex mechanism,the harsh environment and the difficultly detectd key parameters,hence its optimization process is always a research hotspot.In this study,the burden distribution matrix of blast furnace is taken as the research object,the optimal setting of the burden distribution matrix is taken as the goal.The clustering model of burden distribution matrix is established to analyze the historical charing pattern.The data-driven modeling algorithm of burden distribution process in blast furnace is improved to provide the foundation of subsequent analysis and optimization.The classification model of burden distribution matrix adjustment is establishded to judge the burden distribution whether needs to be adjusted.The generation mechanism and optimization model are established to obtain the mapping relation and further optimize the setting strategyof burden distribution matrix,which can improve the setting method of burden distribution matrix.The main contents and contributions are as follows:(1)The clustering model of burden distribution matrix is established.In order to analyze the historical burden distribution matrix of the blast furnace,based on burden surface information and condition parameters,the clustering model of burden distribution matrix is established.The panel data is introduced to analyze the blast furnace production data from time dimension,sample dimension and index dimension.Quantum Clustering(QC)algorithm is studied and used to establish the clustering model.The blast furnace production data are used to testify the clustering model.The high clustering accuracy is obtained,and the feasibility and accuracy of the model is testified.(2)The data-driven modeling algorithm of burden distribution processin blast furnace is improved.Aimed at Multi-layer Extreme Learning Machine(ML-ELM),based on the characteristics of production data,a variety of improved ML-ELM algorithms are proposed to improve multicollinearity,over-fitting and weight selections,generalization ability and accuracy of ML-ELMfor industrial data-driven modeling are improved.The regression models are established,which obtain the better regression results.The feasibility and effectiveness of the improved algorithms are testified.The improved algorithms provide the foundation for subsequentresearch of burden distribution matrix modeling.(3)The classification model of burden distribution matrix adjustment is established.In oreder to judge whether the burden distribution matrix needs to be adjusted,based on the classification algorithms,the operation experience and the production data,the classification modelis established,which obtains better classification results and generalization performance.The feasibility and effectiveness of the classification model are testified.(4)The generation mechanism and optimization model are established.In order to obtain the burden distribution matrix according to the burden surface information,the burden surface parameters are used as input parameters,the number of charging circles for burden distribution matrix is used as output parameter,the neural networkalgorithm is used as the modeling algorithm,and the generation mechanism model of burden distribution matrix is established.To optimize the setting of the burden distribution matrix,the optimization strategy is proposed,and the optimization model of burden distribution matix is established.The production data are used for the simulation experiments to testify the feasibility and effectiveness.The generation mechanism and optimization model of burden distribution matrix improve the the setting method of burden distribution matrix and promote the intelligent development of burden distribution process in blast furnace.