Aplicação do processo de KDD a um ambiente industrial

• Ano de defesa: 2007
• Autor(a) principal: Lucas Costa Oliveira Santos
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Engenharia Elétrica; Engenharia elétrica
• Link de acesso: http://hdl.handle.net/1843/BUOS-8CDHRL

Knowledge extraction from large databases is a complex process which can imply in very high costs, depending on the problem and on what one wants to get. Nowadays, the amount of data stored in many organizations systems goes far beyond human ability to manually interpret and understand that information. In order to deal with this problem,the research area known as Knowledge Discovery in Databases, or KDD, has been created in the Computer Science field.This project, which was motivated by the short exploration of KDD in the Process Industry environment, shows a complete application of this methodology with real data of a Hot Rolling Mill plant in a large Brazilian Steel Industry. Beyond the KDD process presentation, with the definition of every step, this work also reviews the state-of-the-art of this methodology application and of the Data Mining techniques in the Steel Industry, and more specifically in the Hot Rolling Mill. From a group of potential problems, the project main target was defined as theidentification of variables that could be somehow related to the Hot Rolling Mill process Force Error. The CART algorithm comes as the main tool for Data Mining, and its usage resulted in valid and potentially useful discoveries to that Steel Industry, as the correlation between the plant operator actions and the increase of the Roll Force Error,as well as the influence of the Bending Force. Besides the project results analysis, the difficulties found and the near future perspectives of this subject in the Process Industry are presented.