Research And Application Of Machine Turnout Prediction Model Based On Association Rules

An enterprise may accumulate abundant production data during long-term production. So, technicians and quality managers have been concerned with making the best use of data to improve the management and enhance the enterprise's benefit and competitiveness.As a kind of optimum management and enterprise's data solution which ofers decision supports, the data warehouse technology has already been adopted by many enterprises. The core idea of the data mining technology is mining potent, novel rules or patterns used for decision analyses from large amount of data. Based on each machine production processes of Shanghai Baosteel Co,Ltd. Cold Rolling Thin Strip Mill, combined with the development of "Optimized Planning and Evaluation System of Thin Strip Cold-Rolling Process", the modeling technique based on data warehouse and data mining is studied, and the throughput prediction model is established for cold-rolling process.An improved incremental algorithm based on traditional incremental algorithm is proposed, which can mine both strong rules and trend rules and improve the integrity of rules mined. The data warehouse is established after the requirement analysis, design of granularity, design of concept model, and design of dimension table and fact table. The historic data of database system of the enterprise is abstracted, transformed and loaded into the data warehouse. The turnout prediction and trend prediction based on the historic production data and expertise experience is realized through the turnout prediction system, which is developed based on the data of warehouse and expertise experience using the improved incremental algorithm.The turnout prediction system is working in the Cold Rolling Thin Strip Mill, and the prediction precision satisfies the requirement, and the management of production is improved as well.