Blasst Furnacce Coal Injection Rules Mining Based On Data-driven

Blast furnace is a typical complex industry control process with the characteristics of multi-variable, strong coupling, nonlinear and large lag. Blast furnace control process is complex and random that influenced by many factors. It’s not a simple one or a group of parameters of linear or nonlinear feedback control system, it is difficult to use mechanism analysis system to establish the mathematical models to optimize and control the blast furnace operation. The operation of the blast furnace process decision-making depends largely on worker’s experience. Due to the lack of experience, furnace condition is complex and lack of understanding the blast furnace process, making the blast furnace process conditions large fluctuations and coke consumption higher and raises the production costs.Blast furnace process is dependent on the coke. With the continuous expansion of the blast furnace scale in our country, increasingly reducing coking coal reserves and metallurgical coke supply growing tensions.Global iron ore trade is dominated by a small number of exporters and countries, iron ore, coke prices continue to rise, and steel enterprise cost pressure is increased. According to the national12th five-year plan on large enterprise policy of energy saving and reducing energy consumption and serious environmental problem facing our country, each big steel companies are struggling to seek measures of reducing energy consumption. Most enterprise have used pulverized coal injection as an important technical instead of coke and strengthening smelting process in blast furnace.Data mining attracted many scholars’ research and attention in recent years. A lot of production process data stored with the gradual improvement and application of blast furnace automation and information technology. These data contain blast furnace operating rules, the manual operation experiences and detailed reflect the relationship between technical parameters and operating rules and optimized operation mode that useful information for operating decisions and optimal control. But due to the limitation of knowledge acquisition and data analysis ability, make a lot of production data have been idle, which cannot dig up the implicit knowledge. Therefore, on the premise of security, stability and successfully, with low-carbon iron making as the goal, Use the data mining methods seek relationship between pulverized coal injection quantity and coke rate, through reasonable oxygen-enriched pulverized coal injection to achieve the purpose of reducing coke rate, make the blast furnace in the aspect of low consumption, high efficiency, high quality, long life to achieve new breakthrough that has important theoretical meaning and great value of application. This paper is mainly on the research and exploration as follows:1. Basic analyzes the main characteristic of blast furnace smelting process and the optimization decision problems. Presents the basic data mining framework of the blast furnace smelting process, regulate the definition of blast furnace smelting process data mining, basic tasks, implementation process and algorithm to constitute. It emphasizes the basic implementation method and process of coal injection rules mining of blast furnace.2. Using the original data-driven method and based on rough set theory to mining the quantity of coal injection for blast furnace smelting process. Using daily data to establish a global decision-making system, and then through the analysis to break it down into two of fully consistent decision-making system and completely inconsistent decision-making system. For the fully consistent decision-making system, using the face of decision attribute method of abstracting rules, finally conclude the rules of coal injection of blast furnace smelting.3. Use a large number of historical production data of blast furnace smelting, establish the optimization model through optimization methods based on intelligent models (non-precision mathematical model), including fuzzy model, multiple support vector machine (SVM) model, in order to obtain the relationship between the optimization object and the operating parameters, thus achieve the optimization of operating parameters for blast furnace coal injection.4. Studied the method of operation pattern matching and evolution of the quantity of coal injection in blast furnace process. We established a fine coal injection operating mode libraries for blast furnace. Improve the operation speed and accuracy of the pattern matching based on the operating mode of Mahalanobis distance similarity measurement criteria. 5. Use of computer technology and advanced rules mining technology established the operation rules of blast furnace coal injection and fine coal injection operation pattern library based on the data driven. Those can accurately judging the blast furnace condition and control to help master complete the decision-making and optimize.