Efficient Solution for Detecting and Grading Surface Wear on Hot-Rolling Work Rolls

Strategic Insights -

Research conducted by Yanshan University has resulted in an efficient and accurate solution for detecting and grading surface wear on hot-rolling work rolls. The study established an apparatus for capturing high-precision roll surface images and developed a quantitative assessment of common surface wear morphologies. A dataset was constructed, and a MobileNetV2 convolutional neural network (CNN) was employed to develop a grading model that achieved high-speed and accurate grading. The model was able to classify surface wear conditions with minimal model parameters and size, meeting industrial efficiency and practicality requirements.

Key Takeaways:

  • The study aimed to address the issue of wear assessment for hot-rolling work rolls, which significantly affects production efficiency and product quality.
  • The researchers established an apparatus for capturing high-precision roll surface images and developed a quantitative assessment of common surface wear morphologies.
  • A dataset was constructed to support the development of a grading model, which was implemented using the MobileNetV2 convolutional neural network (CNN).
  • The MobileNetV2-wear detection and classification (WDC) model achieved high-speed (21.92 ms) and accurate (96.86%) grading with minimal model parameters (2.27 M) and size (27 M).
  • The model was evaluated against mainstream CNN models, revealing its superiority in terms of efficiency and practicality.
  • A visual analysis of the model classification errors was conducted to identify paths for further optimization.
  • The research provides an efficient and accurate solution for detecting and grading surface wear on hot-rolling work rolls, enhancing product quality and extending the lifespan of rolls.
  • The study was financially supported by the Central Government Guide Local Science and Technology Development Fund Project.
  • The research team consisted of Huagui Huang, Qiwei Hu, Biao Xu, Jiali Zheng, Shimin Xu, and Xinyi Ren, affiliates of Yanshan University and the National Engineering Research Center for Equipment & Technology of Cold Rolling.

Statistics:

  • 96.86% accuracy achieved by the MobileNetV2-WDC model in grading surface wear conditions.
  • 21.92 ms: high-speed grading time achieved by the MobileNetV2-WDC model.
  • 2.27 M: minimal model parameters used by the MobileNetV2-WDC model.
  • 27 M: minimal model size used by the MobileNetV2-WDC model.
  • 2025: year in which the research was published in the Chinese Journal of Mechanical Engineering.

Sources:

  • Machine Vision and Deep Learning for Enhanced Grading and Classification of Surface Wear On Hot-rolling Work Rolls. Chinese Journal of Mechanical Engineering, 2025;38(1).
  • NewsRx. Studies Conducted at Yanshan University on Mechanical Engineering Recently Reported (Machine Vision and Deep Learning for Enhanced Grading and Classification of Surface Wear On Hot-rolling Work Rolls). Journal of Engineering. October 20, 2025; p 3815.