To satisfy the requirements on accuracy and efficiency of point cloud registration of damaged compressor blades a algorithm for partition based on features of neighborhood points and the corresponding accurate point cloud registration of aero-engine damaged blades was proposed. First of all, based on the covariance matrix, a multi-step partition model was employed to define the method to divide feature sub-blocks, and thus obtain effective feature regions. Secondly, a stable n-dimensional feature vector was constructed in accordance with the local curvature, the maximum distance between points, and the angle property of the maximum normal vector; then, by introducing the iterative closest point theory, the minimum Euclidean distance between the corresponding points and that from the point to the surface between the corresponding blocks were established. The accurate position correction of the two models was realized. Finally, the unit quaternion algorithm was used to complete the accurate point cloud registration of damaged blades. Experimental results show that the proposed algorithm can achieve point cloud registration on the surface of the point cloud model of damaged compressor blades, significantly improving the efficiency and accuracy of registration. Moreover, the advantages and robustness of the unit quaternion algorithm are verified through the point cloud database of multiple groups of aero-engine damaged blades.

In response to the issues of poor real-time performance and low detection accuracy when deploying deep learning models for aero-engine damage detection on embedded devices, this paper introduces the FDG-YOLO lightweight model for aviation engine damage detection. Firstly, FasterNet was introduced to restructure the backbone network of YOLOv5, addressing the issue of large parameter count in the backbone network. Second, depth-wise separable convolutions were used to eliminate superfluous parameters in the neck network of YOLOv5 by improving ordinary convolutions. In order to improve the model's expressive power and receptive field, the original C3 structure was replaced with the GS C3 structure, which was built concurrently based on GSConv. Finally, experiments were conducted and validated on an aviation engine damage dataset. In the end, experiments were conducted and validated on an aero-engine damage dataset. The findings show that the FDG-YOLO model reduces the number of parameters by 52.5% and the giga floating-point operations per second by 66% when compared to the original model. On embedded devices, the mean average precision (mAP) reaches 89.6%, surpassing other lightweight models. The frames per second achieves 61, making the detection speed suitable for the engine damage image acquisition rate. It more effectively satisfies aero-engine damage detection's intelligent application criteria.