A Novel Convolutional Neural Network-Based Insulator Defect Detection Method for High-Voltage Transmission Lines

Missing or defect of insulators in high-voltage transmission lines can lead to critical fault of the entire transmission system. The traditional manual patrol method is too inefficient to meet the actual detection needs. Insulator condition monitoring and defect detection of high-voltage transmission lines based on aerial images from unmanned aerial vehicles (UAVs) has been widely investigated. In this paper, an insulator defect detection method based on deep learning technology is proposed using UAVs in high-voltage transmission lines. A staged feature extractor based on the reuse of convolutional feature layers by combining a multiscale feature fusion network is developed. The reuse of convolutional feature layers can reduce the number of channels in the feature map, thus reducing the number of parameters and running faster. Meanwhile, the semantic features can be better extracted by using the features of different receptive fields. The results of training and testing on the Chinese Power Line Insulator Dataset show that our method has higher detection accuracy. The experimental results show that the method proposed in the paper satisfies the insulator defect detection in high-voltage transmission lines, and dramatically improves the detection efficiency comparing with existing methods.