Other events

Title

DIAGNOSIS SOLUTIONS OF ROLLING CONDITION AND PRODUCT QUALITY FOR METAL PLANTS

Authorship

Inami, Haruki; Ishida, Takashi; Imanari, Hiroyuki; Ohara, Kazuhiro

Inami, Haruki

1 Articles, Downloaded 149 Times

Ishida, Takashi

I am this author

Imanari, Hiroyuki

I am this author

Ohara, Kazuhiro

I am this author

DOI

10.5151/9785-9785-32413

Downloads

Abstract

The highly competitive business environment of the metals industry in the world is moving to the demands for more cost effective operation of steel making facilities, rolling mills, and strip processing lines. Achieving stable operation is vital for the steel industry to produce high quality products constantly and productively. In recent years, advances in computer processing speeds, data acquisition rates, and data storage capacity gradually enable industrial applications using new technologies of Big Data, analytics, IoT, and AI for control system designers and process engineers. “Smart Rolling Mill” in the hot strip mill is proposed by our group, and the goal of “Smart Rolling Mill“ is a sophisticated, state of the art, and automated steel rolling mill. It has various solution engines that improve process stability, increase the performance of mill equipment, and provide improved process control. These solution engines are based on extensive knowledge of the rolling process and associated control system. Especially, among these solution engines, the diagnosis system for rolling condition and product quality contributes to cost effective operation and minimizing of yield loss. In this paper, we focus on especially the diagnosis system for rolling condition and product quality in the hot strip mill. Advanced information technologies using Big Data analytics and Machine Learning are applied to real-time data collected from automation system which controls operation and process of rolling. We have developed diagnosis solutions of rolling condition and product quality. The solutions contains predictive and clustering diagnosis. Predictive diagnosis is able to detect a change in the state of facilities or rolling condition, and to avoid a risk of serious trouble. Clustering diagnosis is able to classify fluctuation patterns in the trend chart of product qualities, and to provide effective tuning guidance in order to improve the product qualities. The paper describes these applications with several examples.

Keywords

Diagnosis system; Product quality; Big data; Machine learning