Effective anomaly detection in the steel mill:

The timely detection of faults and anomalies plays an important role in the production of steel. Particularly in the cold rolling of steel sheets, faults and unplanned interruptions can lead to strip breaks, which are associated with high downtimes and costs. In practice, however, it can be extremely difficult to detect a strip breakage in good time and diagnose its cause.
The prerequisite for detecting these faults at an early stage is the reliable and continuous recording and analysis of process data. To do this, process data such as voltage, current or torque must be recorded, evaluated and interpreted in high resolution. However, the amount of data generated during steel strip production is immense. In order to automate the analysis process and keep pace with the constantly increasing requirements, a steel manufacturer is relying on an innovative approach with artificial intelligence. A machine learning solution from time series AI provider Falkonry is trained with historical process data and ongoing production is monitored online using the model to detect signs of a strip break in good time. The iba system provided easy access to the needed high-resolution data for their AI model.