Fault Identification of Transmission Line Shockproof Hammer Based on Improved YOLO V4

Abstract

Anti-vibration hammer, as a key fitting to suppress the periodic vibration and galloping of transmission line wires, plays a very important role in the safe operation of transmission lines. This paper takes UAV aerial images of transmission lines as the research object. Aiming at the small target characteristics of the anti-vibration hammer, a transmission line anti-vibration hammer fault identification algorithm based on the improved YOLO V4 model is proposed. First, this method merges the globalized information obtained after expanding the receptive field with the refined local information. Secondly, use multi-scale convolution kernels to obtain more refined local features, and then use different-scale hollow convolution layers to increase the receptive field and obtain more global data. Finally, the obtained information of different scales is fused. The comparison test with the original model proves that the improved model has a significant improvement in the accuracy of detection.