Back to Webinars
webinars detail page

Trueflaw webinar - Smart NDE

The #AI for #NDE webinar series continues! Join us next week, 2022-09-01 16:00 EEST, to hear Iikka Virkkunen present on smart NDE. Read the intro below and register!

Visit Website

The #AI for #NDE webinar series continues! Join us next week, 2022-09-01 16:00 EEST, to hear Iikka Virkkunen present on smart NDE. Read the intro below and register!

AI has proven a valuable tool to aid in NDE data analysis. In the first instance, the AI models were used to find flaws from the vast data provided by modern NDE techniques. This is still the biggest application, but now we are starting to see a more comprehensive pattern emerge. The solutions no longer use just one model to do one thing. Instead, most of the current systems make use of several specialized AI models that together make sense of the data. One model might detect various areas of interest, and another then focus on some of these areas for specific flaw types. One model might pinpoint the flaw locations and then hand these over to another model that proceeds to run more detailed evaluation on the surrounding data to further characterize and quantify the findings. A model might detect component surfaces to form basis for measurement and another find flaw indications. These combine to provide an informative view for the inspector that provides meaningful quantitative information for evaluated. 

While for now, these are mostly isolated cases, they are starting to connect. As they do, what emerges is a new, smarter NDE, where the data is automatically parsed by range of collaborative models to provide a rich interpretation for the inspector, complete with measurements of interesting features and wealth of additional data to be presented as needed. The NDE gets "smart" in the sense that it can operate on a higher level of abstraction – identify components, surfaces, indications and geometric features instead of amplitudes and a-scans. This then allows things like more meaningful comparison to previous inspection of the same component. 

This webinar looks behind the curtain and talks about the trends we see and where AI for NDE is headed.

About Trufelaw

Trueflaw offers cracks and related services for non-destructive testing. Cracks are used for training, method development and performance demonstration. Today cracks are increasingly used to train machine learning systems for automated flaw detection.

With unique thermal fatigue technology, Trueflaw can grow cracks directly into customer components. These cracks offer the best possible representativeness for any inspection and are used around the world and across many industries, such as nuclear and aerospace industries.

Simplified samples provide a cost effective alternative to cracks in real samples. These can be used, e.g., for performance evaluation of fluorescent penetrant testing, eddy current and various other techniques.

It's not always necessary to invest in your own cracked samples. Trueflaw rents cracked samples for, e.g., probability of detection (POD) purposes. Trueflaw also offers the POD evaluation as a service. It's the easiest and most affordable way to know your POD.

Trueflaw develops and offers automated flaw detection systems based on machine learning. These systems are trained with a combination of client data and Trueflaw's virtual cracks to obtain human level performance. The detection systems are tailored for each customer and validated using POD – the gold standard of NDT reliability.

Tree PNG back


Tree PNG back




Tree PNG back


Application Notes