• 목록

Small steel bars (SSBs) are common steel products manufactured through a rolling process. Surface defects on SSBs, which are critical quality problems of SSBs, are often incurred during the rolling process. In order to prevent the surface defects, key process variables affecting the surface defects should be identified and controlled. Recently, it has been common to collect a large amount of operational data from the rolling process which might include valuable information for improving the rolling process. This study attempts to obtain an optimal condition of rolling process variables where the surface defects are minimized by using a particular data mining strategy called classification and regression tree (CART). By analyzing node information from CART, which includes both leaf nodes and their ancestors, it is possible to identify several candidates for the optimal condition. These candidates are represented in hyperrectangle forms, demonstrating this approach’s robustness and potential for substantial improvements in minimizing the surface defects.