Gwangyang Steelworks Long Product Equipment Department Establishes 'Continuous Casting Equipment Fault Prediction System'

Predicting Core Equipment Failures Related to Quality Using AI and Sending SMS: Advancing from Production and Quality Management to Smart Equipment Management... A Step Forward to a 'Smart Steelworks'

[Asia Economy Honam Reporting Headquarters, Reporter Heo Seon-sik] POSCO Gwangyang Steelworks (Director Kim Ji-yong) Wire Rod Equipment Department announced on the 24th that it has established a ‘Continuous Casting Equipment Failure Prediction System (hereinafter referred to as the Continuous Casting Failure Prediction System)’ applying AI, creating a safe and efficient equipment management environment.

The ‘Continuous Casting Failure Prediction System’ is an integrated equipment management system that combines △data integrated management △real-time equipment status monitoring △AI failure prediction technology. In April this year, the Wire Rod Equipment Department of Gwangyang Steelworks successfully developed it independently and applied it to the 2nd continuous casting plant.

Continuous casting is a process in which molten steel produced in the steelmaking process is poured into a mold and continuously drawn out, cooled with water, and semi-finished products such as slabs are manufactured.

Through this, Gwangyang Steelworks has been able to efficiently manage on-site equipment data. Each equipment had separate condition monitoring systems (CMS) and was managed individually, but now they have been integrated into the ‘Continuous Casting Failure Prediction System.’ In addition, by reflecting past failure case analyses and employees’ equipment management know-how, a real-time equipment status monitoring system was established.

In particular, the AI-based anomaly detection technology was applied to the core equipment of continuous casting, securing differentiated competitiveness and attracting attention. First, AI analyzes the data patterns of normally operating equipment through machine learning. If an anomaly occurs compared to this, it is immediately detected and notified.

Using this system, equipment failures can be predicted up to 72 hours in advance, and messages are sent to the equipment managers via text message, enabling early risk detection and action. This is significant in that it contributed to the prevention of safety accidents using smart technology.

Furthermore, through improved efficiency in equipment management, employees’ working methods have greatly improved. Compared to the previous method of going out to the site to check each piece of equipment one by one, the prediction system first comprehensively identifies the status and then inspects only the equipment with abnormalities. This reduces unnecessary inspections and allows equipment managers to focus on specialized inspections and equipment advancement.

The Wire Rod Equipment Department of Gwangyang Steelworks plans to complete the incorporation of equipment management know-how into the ‘Continuous Casting Failure Prediction System’ by June this year, reduce routine inspection items, and expand specialized inspections. Based on the efficiency and advancement of equipment management, it is expected to realize ‘zero unexpected failures’ and smart working methods.

Han Young-min, leader of the Continuous Casting Maintenance Section at Gwangyang Steelworks, said, "The continuous casting failure prediction system has made it possible to predict equipment abnormalities that were difficult until now, improving not only stable equipment management but also the safety and work-life balance of maintenance staff. We will continue to develop predictive maintenance technology to establish a virtuous cycle system for equipment management."

Meanwhile, Gwangyang Steelworks has selected three plants including continuous casting as pilot plants to establish a smart equipment management foundation, and plans to further develop the integrated equipment management system by packaging the ‘Continuous Casting Failure Prediction System’ and expanding its application to other equipment with the goal of smartifying the entire steelworks equipment.