Prioritized Anomaly Management in Steel Production Using Self-Supervised AI

With the increasing adoption of artificial intelligence (AI), manufacturers are now able to analyze large volumes of data to detect operational anomalies as they happen. However, with limited human attention available to review these anomalies, timely prevention of failure remains a challenge. This paper presents a novel anomaly detection system capable of automatically detecting and prioritizing anomalies that need immediate human attention. Using severity persistence, signal importance and other parameters, the anomaly detection AI enables prioritization of critical anomalies, leading to timely diagnosis and corrective actions. Practical cases of deploying this in commercial operations are presented.

Rishabh Shah | Falkonry
 

Suhas Mehta | Falkonry