AISTech 2022

Sessions Speakers Exhibitors Products Press Releases Show Specials Floor Plan Buzz

Anomaly Detection of the Continuous Casting Strand Guide by Monitoring the Withdrawal Force Using Machine Learning (Room 316)

In the present work, a methodology for detecting anomalies in continuous casting was developed based on the monitoring of the total strand withdrawal force. The anomaly detection is based on the comparison of the actual total withdrawal force with a reference value, calculated by an artificial neural network, which was trained and optimized using the most relevant casting parameters of actual production data. Using this methodology, process or machine anomalies, which call for a visual inspection and therefore a machine downtime, can be detected more rapidly and therefore product quality problems or extensive machine downtime may be avoided.