POSCO Integrates AI in Fault Diagnosis to Reduce Recovery Time by 30%

POSCO employees are accessing the POINT system to learn how to handle abnormal situations. Photo by POSCO

[Asia Economy Reporter Choi Dae-yeol] POSCO has decided to utilize artificial intelligence (AI) technology to repair issues that arise in control devices at its steelworks. The company expects to reduce the average time required to restore equipment by about 30%.

On the 16th, the company introduced the smart search system POINT, which systematically organizes various manuals, technical documents, maintenance techniques, and troubleshooting know-how, developed jointly by equipment control experts and AI specialists. It incorporates AI specialized in natural language processing to enable easier searching.

In the past, when equipment had problems, the response methods were simply listed. The new system predicts the cause of failure by analyzing failure history and correlations with similar cases from other plants, and also provides information on failure probability and inspection priority. POINT is an acronym derived from the English words Posco, Artificial, Intelligence, Trouble-Shooting.

Common control devices at the steelworks include inverters that regulate motor speed and torque, drives, and PLCs that integrate control of these drives and hydraulic devices. They store data from operating equipment and control processes. Due to the variety and complexity of failure types, it is difficult to identify causes, and recovery is challenging because response varies depending on the operator's skill level.

After using the POINT system, feedback on response results and satisfaction improves its accuracy. As usage increases, the diagnostic function is expected to gradually improve. It has also been developed as an application, allowing use on tablets and smartphones.