Fortunately, a sufficient amount of data was available for us to implement an AI-based solution that could address the problem. Cracks Preventer resolves quality issues in casting processes by merging SMS group`s unique domain knowhow with artificial intelligence methods. It prevents downgrading caused by multiple surface defects, like corner cracks, slivers or lamination. To do so, Cracks Preventer learns from historical process data and investigates all relevant process data in real-time. That way defects can be detected before they occur. In this case, we obtained additional real-time data from the customers IBA system, as well as post-process slab data and optical inspection data from their Parsytec system.
Based on the analysis results, Cracks Preventer can predict what is to be done. It presents countermeasures to the operator which then can be implemented into the level 2 process control right away. Thus Cracks Preventer not only avoids downgrading it also helps to smoothen casting processes. Both aspects are of high importance for successively creating a more sustainable metals industry.
In a next development stage Cracks Preventer can be enabled to feedback adapted set-points based on real-time production data into the automation system. That way the necessity for human intervention is reduced even further and product quality in casting processes is improved.