Research on UAV high precision autonomous intelligent detection and evaluation system for large crane equipment metal structure

In order to solve the problem of remote visual detection of inaccessible parts of high-altitude structural parts of large lifting equipment, the thesis designs the upper/lower mounted UAV intelligent detection system for large cranes. The automatic inspection method based on the full coverage inspection path planning and fixed point cruise is proposed to realize the multi-directional detection of key parts without dead corners. The intelligent multi-scale complex defect recognition model based on improved YOLO neural network is established to realize online accurate identification of defects such as cracks, corrosion, coating deterioration, bolt shedding, deformation, wire rope breakage, rust, wear, and deformity.The detection accuracy reached 94.31%, the detection speed reached 17 FPS (frame /s).The crack/corrosion parameter identification and quantitative evaluation algorithm based on the block image segmentation of the defect target was established, which improved the sensitivity and identification accuracy of the structural defect detection. The 5G network is used to transmit images to the server, and the intelligent inspection software for large equipment drones is developed.It achieves accurate online identification of various defects and rapid generation of inspection reports, improving inspection efficiency and accuracy.