Research On Hybrid Algorithm Of Differential Evolution And Support Vector Machine For Energy Prediction In Iron And Steel Industries

Iron and steel enterprise production is featured by high energy consumption,high pollution and high emissions,which exert a serious negative effect on the enterprise's profits and cause environment pollution.In the face of some problems hindering the development of iron and steel enterprises,such as energy medium coupling,large energy consumption,low energy efficiency,scientific and reasonable method of energy management is in urgent need for enterprises to reduce the energy emissions,improve energy utilization efficiency and realize energy recycling.Energy prediction problem is studied in this thesis,and least square support vector machine(LSSVM)algorithm is used to predict the energy consumption with different time dimensions so as to better monitor the energy medium consumption and regeneration,which consequently can help to reasonably production,reduce energy emissions,and improve energy recovery rare.The main contents are as follows:(1)In current iron and steel enterprises,there are many problems in the process of making energy plan,such as complex energy consumption and regeneration,a long time needed for energy planning,and great influence of artificial factors.Based on these,the thesis classifies the energy prediction problem in the view of the dimension of time and medium,and analyzes the main factors that have significant influence on energy consumption of main procedures through technological mechanism so as to lay a foundation for energy management.(2)The production and energy consumption process in iron and steel industry is complex due to unknown mechanism model and nonlinear energy conversion,so the LSSVM is used to construct the prediction model,and controlling variables method is designed to manually set the unknown parameters that affect prediction precision.To verify the validity of the prediction model,LSSVM algorithm with two different kernel functions is used to make the prediction based on the blast furnace production data.(3)LSSVM algorithm is used to deal with the problem of energy medium prediction in iron and steel enterprises and uses differential evolution(DE)is adopted to optimize the parameters of the prediction model.To demonstrate the efficiency of DE,the results obtained by hybrid algorithm are compared with that obtained by the particle swarm optimization algorithm.Based on the real-time blast furnace production data and steelmaking month data,the computational results show the hybrid algorithm of DE and LSSVM has a higher prediction precision.(4)With the hybrid algorithm embedded,the energy medium prediction system for iron and steel enterprises is designed and developed in which the database structure and forecast system function modules are designed.This system provides helpful support for iron and steel enterprises to establish the energy plan and realizes energy prediction according to different time dimensions.The main functions include basic data management,parameter setting,energy prediction and system management.