Speaker
Description
Detailed inspections of aircraft bodies are designed to detect and characterize damages, whether accidental (impact) or progressive (mechanical fatigue, ageing), and affecting the integrity of the aircraft structure. These inspections, mainly non-destructive tests (NDT), are carried out as part of planned maintenance operations or following the detection of damage. They are time-consuming, requiring the dismantling of parts of the aircraft to access the inspection zone, or the use of lifting equipment to access a high part of the aircraft. Automating these tasks using ad-hoc inspection systems is a very attractive solution for reducing the time and cost associated with these inspections. In this context, we developed a multi-sensor robotized NDT platform coupling two inspection modalities: visible light for surface damage detection, and infrared light for in-depth damage detection. Infrared detection is based on active thermography, in which the robotic arm carries both the imaging system and a thermal exciter. Acquisition is performed in line-scan mode, by heating the sample at constant speed, angle and distance to the arm before the IR camera collects the thermal response. In this way, each area of the sample is guaranteed to be scanned with more uniform sampling conditions. Compared to classical static flash thermography, this extra degree of freedom makes it possible to inspect parts with considerable curvature, such as aircraft fuselages and nacelles. Internal damage of such parts can be accessible with better confidence. The 3D cartography of the composite part is recorded by a depth camera and used to calculate the optimum scanning trajectory for the inspected parts. The thermal and visible field is then provided ad-hoc with damage localization. A detailed characterization of the multi-modal robotized NDT platform developed at ONERA was carried out to evaluate its ability to reliably detect internal defects in aeronautical composite structures. A sensitivity study was conducted in line scan thermography operating mode to determine the detection limit in a CFRP composite plate with calibrated detects. The integration of advanced thermal data processing algorithms, particularly variational auto-encoders algorithms, has proven to significantly improve internal defect detectability and robustness to imaging constraints, such as defocus. A multi-parameter study on a lightning-struck composite plate with real internal damage was conducted to optimize line-scan inspection, supported by a parametric simulation study. The links between line scan velocity and damage detection sensitivity have been investigated, and line-scan parameters were optimized with regard to the depth of internal damage. Ongoing efforts focus on deploying these methods for damage monitoring in aeronautical applications, utilizing 3D camera-acquired cartography.
