Data Warehouse And Data Mining's Application In Steel Enterprise

Long-term production has accumulated abundant data about hot-roll product' s performance in one steel enterprise. A steel enterprise aims to improve her economic benefit and enhance her competition. So, technicians and quality managers have beening wanting to making the best use of data to enhance the product' s performance and design new products and new technics all along.The dada mining technology based on the data warehouse was prisented at the beginning of 90's in 20 centuries.It is the core idea to guide the technology decision of product' quality and come out enormous economic benifits by mining effective, novel, potent avail, understandable rule or pattern from 'data ocean' .In this paper, we establish the data market on hot-roll product's performance by mature data warehouse and modeling. Because of the complexity of data origin, we must extract, convert and load the needed data, namely ETL process. Then the processed data is integrated to the data market on hot-roll product' s performance.Two kinds of model based on BP neural network and multi-line regression are established by the data market on hot-roll product's performance. The two models' s precision are 84.7% and 84. 3%separately. They meet customer' s desired presion. Compared with the latter, the former is better. Then we can design new products and new technics by the established model. Durning the experiment, average analysis help to confirm variable scope. Then relativity analysis analyze the infection between input variables and output variables. Principal factor analysis evaluate the chemical elements synthetically to get the most information.The paper' s models have applied to one steel enterprise. They have enhanced hot-roll product's performance, and have strengthened the competition. So the enterprise obtain great economic benefit.