English：

To address the timeconsuming, laborintensive, and inefficient nature of manual inspection of welded joints in reinforcement cages, this paper proposes an intelligent detection technology based on object detection algorithms and fuzzy logic reasoning. Firstly, an improved YOLOv5 algorithm was used to capture images of the welded joints and extract features, with the introduction of a RepVGG module to enhance detection efficiency. Next, by integrating fuzzy logic reasoning, a fuzzy set of joint features and a fuzzy relationship matrix were constructed to comprehensively evaluate multiple defects, such as warping and volumetric shrinkage. Practical results demonstrate that this method significantly improves detection accuracy compared to the standalone YOLOv5 model, providing effective technical support for the automatic inspection of weld joint quality.