Exposed conductors in 10 kV distribution line are one of the major causes with operational faults in distribution lines

continuously affecting the safe and stable operation of the distribution network. Traditional manual inspection methods often fail to detect such defects in a timely manner. A detection method for exposed conductors in 10 kV distribution lines is proposd based on an improved YOLOv8 algorithm

which is designed to assist power grid maintenance personnel detecting conductor exposed defects quickly and efficiently. The algorithm replaces the original convolution with omni-dimensional dynamic convolution in the backbone network

enhancing the features of exposed conductors through multi-dimensional feature extraction. In the neck network

the connection between high-level and low-level features is enhanced by combining attention embedding module with the cross stage feature fusion module of the original network

thereby analyzing both the overall shape and local details of exposed conductors. For the loss function

distance intersection over union with normalized wasserstein distance is combined to increase focus on cases where targets are small or background interference exists in drone inspection photographs. The experimental results demonstrate that the improved algorithm achieves increases of 4.8 percentage points

4.2 percentage points

and 5.2 percentage points in precision

recall

and mean average precision

respectively

compared to the original algorithm. This effectively enhances the detection capability for exposed distribution conductors

providing a new technical approach for ensuring the safe and stable operation of power systems.