01/1999 - 12/2002
Hi-Tech Steel is a large-scale project, funded by TEKES, and it consists of many subprojects. Our group confines to the data mining subproject, which goal is to survey the state of the art of the data mining globally as well as implement advanced data mining techniques in practice based on this knowledge. New techniques are under study too. This subproject supports Hi-Tech Steel project giving proven techniques and problem solving suggestions to other subprojects.
The data mining subproject contains three closely related workflows: data preprocessing for data mining study, roller campaign information study and furnace control data study. The data preprocessing surveys all the recorded data variables from the rolling process. The goal is to find variables and data items that have a significant contribution into output observed – defective coils. The roller campaign study tries to find answers to question: how does the roll changes effect on the risk for a single coil to be a defective after the rolling process. The furnace control study surveys the dependencies between furnace control variables. The analysis is done using graph theory and neural networks. The goal of the study is to create a furnace control algorithm, which is more accurate than the existing one.