Knowledge and technology transfer through standardization
Despite their relevance, standardization is often underestimated as an instrument for knowledge and technology transfer. Fraunhofer IZFP has recognized both this fact and the general trend of research organizations and industry turning away from participation in standardization committees and has taken up the topic as a strategic focus in the service area “Consulting and holistic services around measurement, testing, data value creation and standardization“. In addition to the cooperation in already established and active DIN standardization committees such as NA 062-08-27 AA “Thermographic testing“, NA 092-00-27 AA “Friction stir welding“ and NA 062-08-23 AA “Ultrasonic testing“ as well as the Europe-wide active CEN/TC Committee138/WG 11 “Infrared and thermographic testing“, the co-design of new standards and qualification procedures in the context of intelligent sensors and sensor networks is to be strengthened in the future. Among other things, AI-based evaluation or decision-making processes will play an essential role here. However, there are currently only a few sets of rules that describe such processes in a normative framework.
NDE 4.0 systems represent a revolutionary and disruptive step for the NDT world of the future. They use machine learning-based algorithms for self-optimization and self-monitoring, and neural networks for data analysis. These technologies are making inroads at tremendous speed into the highly deterministically planned and driven standardization and approval of NDT technologies that had previously grown for more than 40 years. Therefore, corresponding norms and standards for quality-assured use and for verified quality in the industrial introduction of NDE 4.0 systems must be adapted or dynamized qualification processes must be developed and embedded in a normative framework. Fraunhofer IZFP is making significant contributions to solving this challenge in the following subject areas:
- Adaptation and further development of validation and qualification processes as well as documentation for the introduction of new NDT procedures based on the flexible laboratory accreditation according to DIN EN ISO/IEC 17025.
- Conception and creation of reference, test and validation data sets for AI-supported NDE 4.0 algorithms on hardware (firmware, FPGAs and ASIC implementation) and software level (evaluation software with the aid of Machine Learning (ML), Deep Neural Networks etc.) for specific, already classically established application segments in NDT.
- Development of scientific-technical specifications and methods for the verification of AI and ML applications with respect to comprehensibility, trustworthiness and plausibility (“Explainable AI“ or “Trusted AI“).
- Development of process descriptions for the comparison of new test result assessments and evaluations, which are based on AI/ML algorithms (or use them at least partially) with firmly prescribed testing and evaluation standards of the NDT, i.e., with previous specifications in existing object standards.
These activities are also actively promoted in newly founded or revived standardization committees with significant participation of Fraunhofer IZFP. Examples include the standardization committees NA 062-08-29 AA “Virtual sensor“ and NA 062-08-28 AA “Methodology for qualification of non-destructive testing“. Furthermore, the project idea developed at Fraunhofer IZFP “Innovation acceleration through flexibilized validation and certification paths for NDE 4.0“ was included as a lighthouse project in the 2nd edition of DIN's German standardization roadmap for artificial intelligence, which demonstrates the relevance of the topic.