BEGIN:VCALENDAR
VERSION:2.0
METHOD:PUBLISH
BEGIN:VEVENT
ORGANIZER;CN=ESTAD 2023:mailto:info@metec-estad.com
LOCATION:Room 02
SUMMARY:Novel analysis techniques for evaluating iron ore sinter based on artificial intelligence
DESCRIPTION: The construction of a smart factory platform for digital manufacturing is being promoted by international steel corporations. In this environment, we are implementing AI technologies to replace existing analysis methods and reduce human error. This is referred to as a "smart analysis system" by Hyundai Steel and focuses on the analysis of the raw materials used in ironmaking process as well as the evaluation of the quality of the final product. In this conference, I'd like to share two novel analysis methods for iron ore sinter with deep learning and chemometrics. 
 In the agglomeration process, iron ore sinter is produced for use in blast furnace. Sinter quality and strength are related with the iron ore sinter's mineralogy and microstructure. By using an optical microscope, it is possible to identify the main mineral phases in iron ore sinter, which include hematite, magnetite, calcium-ferrite, and slag. To evaluate quality and operate the ironmaking process in relation to fuel cost optimally, it’s essential to evaluate the fraction for each phase. The phase classification and quantification in the field is currently carried out manually by an analyzer using the naked eye. In this study, a new automated analysis method using deep learning is proposed to replace human inspection for mineral phase fraction. It is notable that the automatic labeling method utilizing clustering analysis has significantly reduced the time required for deep learning-based semantic image segmentation.
 One of the parameters for the reduction index and reduction degradation index of iron ore sinter is the magnetite ratio. Every four hours in the field, the magnetite content has been measured using the titration method. The existing method, which requires a lot of time, is proposed to be replaced by a new analysis approach that employs Raman spectroscopy and chemometrics.
CLASS:PUBLIC
DTSTART:20230614T151000
DTEND:20230614T153000
END:VEVENT
END:VCALENDAR