融合蒸汽参数调节与图像识别的钢带表面缺陷检测方法
By integrating steam parameter adjustment and image recognition, a condensation development pattern is formed using a water-cooled roller and camera array. Defect detection is performed using a deep learning model, and long and short cycle decoupling control solves the problems of inaccurate low-contrast defect detection and the inability to adaptively adjust steam parameters in existing technologies, thus achieving efficient and stable steel strip surface defect detection.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ZHANGJIAGANG YANGTZE RIVER COLD ROLLED PLATE CO LTD
- Filing Date
- 2026-06-16
- Publication Date
- 2026-07-17
AI Technical Summary
Existing methods for detecting surface defects in steel strips are difficult to reliably detect low-contrast defects under complex conditions such as high speed, heavy load, and high temperature. Furthermore, the steam condensation development method cannot be adaptively adjusted, resulting in a high rate of missed detections and unstable detection results.
The method integrates steam parameter regulation and image recognition. It calculates the initial steam parameters by real-time acquisition of steel strip state parameters, uses surface micro-nano textured water-cooled rollers to form condensation and development patterns, combines linear array and area array cameras to acquire images, inputs them into an improved YOLO-Cond model for defect detection, and adjusts the steam and cooling water parameters through long and short cycle decoupling control to achieve closed-loop control.
It improves the detection sensitivity and real-time performance of low-contrast defects, reduces the false negative rate, and achieves stable detection under complex working conditions, meeting the real-time and accuracy requirements of high-speed production lines.
Smart Images
Figure CN122409677A_ABST