The aviation supplier
PFW Aerospace, a Germany-base subsidiary of the French company Hutchinson, is evaluating the use of Artificial Intelligence (AI) to ensure high non-destructive testing quality standards.
VisiConsult, a manufacturer of customer-specific X-ray systems, has developed a dedicated AI toolbox to automatically evaluate X-ray images. The core of the joint evaluation is the qualification of AI according to demanding aviation test standards.
Markus Gutensohn, head of R&D at PFW Aerospace, said: “When VisiConsult announced that it had developed AI algorithms to automatically evaluate X-ray image data and they offered a pilot programme for this purpose, it was clear to us that we should be involved.”
PFW Aerospace subsequently selected as a pilot customer and will receive an AI prototype including test operation and a statistical qualification report.
The application is for X-rays of longitudinally welded seams on titanium tubes. The test parts are subject to very high-quality standards as they are installed in aircraft world-wide. PFW will collect X-ray images over a defined period of time, which VisiConsult will then use as a basis to train AI networks. The AI itself is operating on secure industrial platform developed in collaboration with Fujitsu.
Demanding inspection standardsIn order to qualify this new technology according to the demanding inspection standards, sufficient statistical data about the accuracy of the algorithms must be available. Therefore, there is a permanent comparison of the AI results with the decisions of experienced NDT inspectors. With the help of this input, the neuronal networks are also regularly retrained. Only when a sufficiently good detection probability (POD) has been proven can the system can be qualified and used in production.
PFW Aerospace sees AI for defect interpretation as a future-oriented tool for its quality department. Mr Gutensohn added: “This will save our staff a lot of time interpreting X-ray images, and this time is ultimately available for other important tasks.”
The use of AI is expected to significantly increase the efficiency of non-destructive testing. It sorts out components with detected deviations and only forwards the necessary X-ray scans to the inspectors for final approval by humans. The information about defect locations and trends can also be sent to the development team in order to optimise design data. Mr Gutensohn continued: “The final quality approval will continue to be done by our highly qualified NDT colleagues.”
For Lennart Schulenburg, VisiConsult managing director, the fact that AI is the next step in quality assurance is a logical consequence of the digital revolution. He said: “Our inspection systems generate a magnitude of digital images and data that has to be evaluated and transformed into information. This amount of data is hardly manageable for humans, but it can be processed with high precision by AI systems. Therefore, we see AI as a helpful tool to assist humans to perform more effectively.”
VisiConsult is planning further pilot programmes in other industry sectors, including oil and gas and automotive. More information can be found at the website (
www.visiconsult.de/smart-inspection).