Intelligent detection and modelling of composite damage based on ultrasonic point clouds and deep learning

This paper proposes a novel composite material detection and modelling method: First, the robotic arm controls the ultrasonic probe to automatically scan the composite material and obtain the ultrasonic point clouds through data processing. Then, the ultrasonic point clouds are intelligently detected by the PVT-RCNN model to get the 3D spatial information of the damage. Subsequently, the damage-free model and its node point clouds are constructed according to the material characteristics, and the Rodrigues Iterative Closest Point (RICP) algorithm is used to realise the registration of the ultrasonic point clouds and the node point clouds. Finally, the damage points cloud features are generated in the model to obtain the finite element model containing prefabricated delamination damage. The results show that this method can conduct intelligent ultrasonic detection of composite materials and directly generate a model with damage, effectively evaluating the material's performance.