Research On Abnormity Monitoring And Diagnosis Of Steel Production And Material Quality

In the production process of iron and steel, molten iron is generated by burning material surface. The distribution of the material surface directly impacts on the quality of the molten iron, thus impacts on the quality of steel. The distribution of the material surface can’t be observed directly in the production process due to its physical shape as well as the high temperature characteristic of blast furnace production, making it difficult to grasp the impact of the blast furnace conditions on molten iron quality. So we research the monitoring and diagnosis of abnormal material surface of blast furnace. It’s important for improving the quality of molten iron, reducing the waste materials and improving the quality of steel products.The production plan of iron and steel enterprises is organized by customer orders. All of the steel materials are matched to the appropriate customer orders. However, some of the steel materials turn into open-order materials as a result of their unsatisfactory quality. It causes the increasing of the raw material inventory and the cost of the enterprise, and even affects the normal production of enterprises. Therefore, it has a very important significance on improving the rate of qualified products and reducing the open-order materials to analysis the causes of abnormal steel materials as well as monitor and diagnose the abnormal conditions.Based on the idea of data analytics, this thesis researches the monitoring and diagnosis problems of the abnormal material surface of blast furnace and the abnormal quality of the steel materials, uses the method of support vector machine (SVM) to monitor and diagnose the reasons of the abnormal material surface of blast furnace and the abnormal quality of the steel materials. For the abnormal material surface of blast furnace problem, an idea of online support vector machine is proposed in the diagnosis method. For the abnormal quality of the steel materials problem, the estimation of distribution algorithms (EDA) is introduced into the diagnosis method to improve the performance of SVM.This thesis main research includes the following parts:(1) Research on the problems of the abnormal material surface of blast furnace under the background of the production process of blast furnace. Analyze the characteristics of blast structure and ironmaking production, and the relationships of the temperature and the material surface in the blast furnace. Then research on diagnosis of abnormal Material surface. Propose the method of online support vector machine because of the tightness, the high temperature resistance and the real-time of ironmaking. Use the real-time data to training the mathematical model of SVM. Diagnose the abnormal blast furnace surface material online with using the mathematical model. The results showed that the diagnosis of the abnormal material surface is accurate.(2) Analysis of the abnormal quality of the steel materials. Analysis the reasons of the customer order’s materials that becoming the open order’s materials. Use the historical data of the steel materials to train the mathematical model of SVM. Use the estimation of distribution algorithms (EDA) to improve the performance of SVM. Diagnose the abnormal quality steel materials with using the mathematical model. The results showed that the diagnosis of the abnormal quality of the steel materials is accurate.(3) In the context of the actual steel production process, Designed and developed the system of the diagnosis of the abnormal quality of the steel materials. It included spot cause analysis, the management of noon-meeting and the contract management. It provided supports of data for decisions of staffs.