One of the major challenges when it comes to the defect classification procedures commonly employed in the manufacturing and other industries is how to increase both reliability and performance. List is using numerical simulation to develop diagnostics that are fast, flexible enough to adapt to manufacturers' situations, and that can be integrated directly into manufacturers' existing measurement systems.

List researchers used models from the institute's CIVA simulation software to develop generic supervised learning tools and meta-models that enable automatic diagnostics that can be integrated into the production cycle. The methods can be used to identify and reliably characterize anomalies detected during inspection. Furthermore, the machine learning techniques used can operate in real time, which means that the system gets smarter and more robust with use, even in the presence of disruptive events in the environment.

The initial results obtained using the techniques were presented at the QNDE 2017 international conference. Potential industrial partners have expressed interest in this rapid diagnostic technique.