Aiming to address the issue of insufficient accuracy in identifying various types of substation equipment in challenging environments such as low light and fog, a novel intelligent recognition method for substation equipment is proposed based on the fusion of visible and infrared images. The image undergoes preprocessing steps including Gaussian filtering and Sobel operator sharpening to enhance its quality. The contour angle orientation (CAO) method is employed for registering visible and infrared images. Furthermore, a gray-weighted average fusion technique is utilized to construct a comprehensive dataset comprising 2384 images from four different types of equipment, which are then used for model training purposes. The trained YOLOv8 model demonstrates remarkable performance in detecting 477 test samples of substation equipment with a mean average precision (mAP) reaching 90.52%. Additionally, a comparison between the proposed fusion-based approach and the single light source detection method reveals that our proposed method achieves higher accuracy levels. This research contributes towards facilitating daily maintenance and overhaul activities related to substation equipment by power operation and maintenance personnel.