ABSTRACT Power systems are pivotal to safeguarding human production and daily life, and switchgear cabinets constitute vital components of substations. Conducting wiring inspections on these cabinets is crucial for ensuring the secure operation of power systems. Addressing the issues of low accuracy and high false detection rates in existing inspection methods, this research proposes a substation switchgear cabinet wiring detection approach that integrates an improved convolutional block attention module with You Only Look Once 8 (YOLOv8). The research innovatively proposes a comprehensive image enhancement strategy combining affine transformations, contrast enhancement, and mosaic occlusion. An improved Convolutional NeXt module, based on masked autoencoders and response normalization, is introduced into the YOLOv8 backbone network to enhance feature extraction capabilities. The standard convolutional block attention module is optimized to reduce information loss and increase focus on key features. Experimental results demonstrate that the enhanced YOLOv8 model achieves a mean Average Precision (mAP@0.5) of 96.5% and precision of 97.2% on a proprietary dataset, processing 158 frames per second. This significantly outperforms baseline models including Faster R‐CNN, YOLOv8, and YOLOv9. Ablation studies further validate the effectiveness of each enhanced module. Consequently, the proposed methodology substantially improves both the accuracy and speed of substation panel wiring detection. This provides reliable technical support for the secure operation and maintenance of power systems, thereby ensuring stable grid functionality.
Lei et al. (Sun,) studied this question.