To implement computer-aided preselection of defects using artificial intelligence algorithms

For space applications, the reliability of any component with respect to mechanical failure is a key requirement, especially in the case of ceramic components, where a single defect at a critical position may lead to crack initiation, growth and fracture of the component under mechanical or thermomechanical load. Instead of proof testing, an automatized and reliable method based on Computed Tomography (CT) non-destructive testing methods supported by a fast data evaluation would reduce ceramic components development risk and cost. Identification and categorisation of defects and assessment of their degree of criticality for fracture, based on the results of finite element (FE) simulated stress field of application conditions is required. Based on a database of CTs of sample components, an AI algorithm has to be developed to automatically identify the defects and to optimise the scanning pattern by improving the resolution only where potential defects are identified.This activity aims to accelerate the CT scan process of ceramic components while providing an increased sensitivity to smaller defects. The outcome of the GSTP activity GT17-320MS "Improvement of NDI-methods for ceramic structures" which included extensive CT scans of Si3N4 ceramic components, are essential for this activity.This activity encompasses the following tasks:- Assess existing ceramics CT scan to identify the potential defects - Explore potential algorithms to improve the detection sensitivity (at the expense of additional false positive)- Develop an AI algorithm to classify the findings between defects and artifacts based on previous ceramic CT scans- Using AI algorithms in the loop, optimize the CT scan strategy (automatically increasing resolution to resolve potential defects)- Feed the CT scan strategy with the results from FE modelling to define the target resolution for each zone