Deep Learning and Transfer Learning Based Detection of Missing Fasteners for Wind Turbine Installation

In the installation and construction of wind power projects, its important fastener bolts bear the responsibility of connecting and fixing important parts, which directly affects the safety, stability and performance of wind turbines. Aiming at the problems of missing nuts caused by various factors and low efficiency of manual detection, a nut missing intelligent detection method is proposed. Firstly, the nut missing image dataset is collected and labeled, and the training set, validation set and test set are divided in the ratio of7:2:1; The Pytorch framework is used to build the YOLOv8 target detection network and train the dataset using the migration learning method. After 300 iterations of training, the accuracy of missing nut detection reaches 89.6%. The results show that the use ofYOLOv8 algorithm and migration learning can achieve accurate detection and can be used for nut missing detection in the construction of real projects.