Weld intelligent defect detection model training method, detection method and electronic equipment
By combining the YOLOv8 object detection network with the PPO reinforcement learning model, an intelligent weld defect detection model was constructed, which solved the problem of high false alarm rate in weld defect detection and achieved high-precision, low-false-alarm automated detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- CHONGQING UNIV
- Filing Date
- 2026-04-21
- Publication Date
- 2026-07-21
AI Technical Summary
Existing weld defect detection methods rely on manual interpretation, resulting in a high false alarm rate. Furthermore, existing automatic detection methods have a high false alarm rate in complex industrial scenarios, requiring manual verification and failing to effectively reduce false positives.
The YOLOv8 object detection network is combined with the PPO reinforcement learning model. By constructing a reinforcement learning environment and using reward calculation rules, the intelligent weld defect detection model is trained to optimize the adaptive decision-making of candidate boxes, delete false positive boxes and retain reliable candidate boxes.
It significantly reduces the number of false alarms, improves detection accuracy and recall, maintains high detection precision and reliability, reduces the false deletion of true cases, and enhances the integrity of detection results.
Smart Images

Figure CN122199516B_ABST