Study On The Data Analytics Of Hot-chain Process In Steel Plant

Hot-chain process,which includes ironmaking,steelmaking,continuous casting and hot rolling process,is very important in the steel production.The prediction and control of the temperature parameters in the hot-chain process is necessary for saving energy and reducing consumption.In this thesis,throat temperature field,which is a key parameter in the blast furnace operation,and hot charging rate,which is a key performance indicator in the steelmaking-continuous casting-hot rolling process are predicted,respectively.The prediction result can provide the scientific basis for production and management,and can also help planners find out problems and solve the problem in time.It can help realize the informatization,intelligentialize,and refine of the steel production and management,and hence improve the profit and competitive power of the steel enterprise.The main works of this thesis are summarized as follows:(1)For the prediction problem of the throat temperature field in blast furnace,taking the statement parameter,operation parameter,and temperature parameter as the initial imput variables,the actual input variables of model are selected for each temperature measuring point by conducting a series data preprocessing that includes abnormal data processing,correlation analysis of influencing factors and uniform dimension.Support vector machine(SVM)is used to establish the cross measurement points’ temperature prediction model,and estimation of distribution algorithm is adopted to optimize the parameters of SVM.Finally,the temperature field of blast throat is built by using the three spline interpolation method.The computation results show that the prediction model built for different temperature measuring points improves the prediction accuracy and has a better tracking performance.In addition,the fitting accuracy of temperature field is no more than 3.92%,which can satisfy the accuracy required by the practical production.(2)For the prediction of hot charging rate,based on the given potential factors in the steelmaking and continuous casting stage,the slab specifications,slab composition,slab production route,slab cutting temperature,slab inventory of slab yard,planning weight of hot rolling,and due date are selected as the initial input variables by using the analysis methods of scatter chart,control chart,regression fitting and correlation analysis.According to the steel grade of slab,the actual input variables are selected by correlation analysis,and then the prediction model is built for each given steel grade.Finally,the charging rate is calculated based on the predicted charging temperature of slab.The computation results show that the average prediction accuracy of hot charging rate reaches 93.45%,which is able to satisfy the actual forecast requirements.(3)Embedding the hot charging rate prediction model,the hot rolling production process performance evaluation subsystem is designed and developed for an iron and steel enterprise.Based on long-term practical investigation and analysis of user requirements,several indicators such as hot charging rate are selected as the key performance indicators.According to the design of system function structure and database structure,the functions of data storage,statistics,analysis and prediction of key performance indicators are realized.The system can track and monitor the hot rolling process and employee performance.