基于深度学习的真空镀膜生产缺陷检测方法研究

doi:10.13385/j.cnki.vacuum.2025.04.13

真空 ›› 2025, Vol. 62 ›› Issue (4): 69-74.doi: 10.13385/j.cnki.vacuum.2025.04.13

基于深度学习的真空镀膜生产缺陷检测方法研究

倪俊^1,2, 郭腾^1,2, 李灿伦^1,2, 侯凯霖³, 李荣义³

1.上海卫星装备研究所, 上海 200240;
2.上海航天裕达科技有限公司, 上海 200240;
3.哈尔滨理工大学先进制造智能化技术教育部重点实验室, 黑龙江哈尔滨 150080

收稿日期:2024-12-02 出版日期:2025-07-25 发布日期:2025-07-24
通讯作者: 李灿伦,研究员。
作者简介:倪俊（1982-）,女,四川绵阳人,博士,高级工程师。

The Surface Defect Detection Method of Vacuum Coating Production Based on Deep Learning

NI Jun^1,2, GUO Teng^1,2, LI Canlun^1,2, HOU Kailin³, LI Rongyi³

1. Shanghai Institute of Spacecraft Equipment, Shanghai 200240, China;
2. Shanghai Aerospace Yuda Technology Co., Ltd., Shanghai 200240, China;
3. Key Laboratory of Advanced Manufacturing and Intelligent Technology, Ministry of Education, Harbin University of Science and Technology, Harbin 150080, China

Received:2024-12-02 Online:2025-07-25 Published:2025-07-24

摘要/Abstract

摘要： 柔性热控薄膜的热辐射性能是影响航天器表面热平衡控制的关键。在真空磁控溅射过程中,由于设备实时工艺波动,薄膜可能会出现多种缺陷。为保证镀膜生产质量与效率,针对传统人工缺陷检测方法依赖经验、效率低、劳动强度大、精度低和实时性差等问题,提出了一种基于高效自适应卷积和通道-空间金字塔池化的镀膜缺陷检测方法,详细介绍了模型各关键模块的优化设计方法。结果表明,该方法可自主学习建模潜在分布的多层表征,通过深层神经网络逐层提取所检测目标的特征;其对8类缺陷的平均识别准确率达到了约93%,识别速度FPS提升至140,可以有效提高产品生产质量,满足批量化、高效率的生产需求。

关键词: 热控薄膜, 真空镀膜, 深度学习, 缺陷检测

Abstract: Flexible thermal control coating is the key for controlling the surface thermal balance of spacecraft. During the vacuum magnetron sputtering preparation process, various coating defects often occur due to real-time process fluctuations of the equipment. The traditional manual defect detection methods have the problems of relying on manual experience, low efficiency, high labor intensity, low accuracy, and poor real-time performance. To ensure the quality and efficiency of coating production, a coating defect detection model based on efficient adaptive convolution and channel space pyramid pooling was proposed, and the optimization design method of each key module was introduced in detail. The results show that this model can autonomously learn and model multi-layer representations of potential distributions, and extract features of the detected targets layer by layer through deep neural networks. The average recognition accuracy for 8 types of defects reaches about 93%, and the recognition speed FPS increases to 140. This method can effectively improve the quality of product production and meet the demand for efficient mass production.

Key words: thermal control coating, vacuum coating, deep learning, defect detection

中图分类号: TB43

倪俊, 郭腾, 李灿伦, 侯凯霖, 李荣义. 基于深度学习的真空镀膜生产缺陷检测方法研究[J]. 真空, 2025, 62(4): 69-74.

NI Jun, GUO Teng, LI Canlun, HOU Kailin, LI Rongyi. The Surface Defect Detection Method of Vacuum Coating Production Based on Deep Learning[J]. VACUUM, 2025, 62(4): 69-74.

参考文献

[1] ZHENG X P, ZHENG S, KONG Y G, et al.Recent advances in surface defect inspection of industrial products using deep learning techniques[J]. The International Journal of Advanced Manufacturing Technology, 2021, 113: 35-58.
[2] WAN X, ZHANG X Y, LIU L L.An improved VGG19 transfer learning strip steel surface defect recognition deep neural network based on few samples and imbalanced datasets[J]. Applied Sciences, 2021, 11(6): 2606.
[3] REDMON J, DIVVALA S, GIRSHICK R, et al.You only look once: unified, real-time object detection[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2016: 779-788.
[4] REDMON J, FARHADI A.YOLO9000: better, faster, stronger[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2017: 7263-7271.
[5] REDMON J, FARHADI A. YOLOv3: an incremental improvement[R]. arXiv:1804.02767, 2018.
[6] BOCHKOVSKIY A, WANG C Y, LIAO H Y M. YOLOv4: optimal speed and accuracy of object detection[R]. arXiv:2004.10934, 2020.
[7] LI C Y, LI L L, JIANG H L, et al. YOLOv6: a single-stage object detection framework for industrial applications[R]. arXiv:2209.02976, 2022.
[8] WANG C Y, BOCHKOVSKIY A, MARK LIAO H Y. YOLOv7: trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]// 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Vancouver, BC, Canada: IEEE, 2023:7464-7475.
[9] WANG C Y, YEH I H, MARK LIAO H Y. Yolov9: learning what you want to learn using programmable gradient information[C]// Computer Vision-ECCV 2024. Milan, Italy: Springer International Publishing, 2024.
[10] WANG A, CHEN H, LIU L H, et al. YOLOv10: real-time end-to-end object detection[R]. arXiv:2405.14458, 2024.
[11] KHANAM R, HUSSAIN M. YOLOv11: an overview of the key architectural enhancements[R]. arXiv:2410.17725, 2024.
[12] LIU W, ANGUELOV D, ERHAN D, et al.Ssd: single shot multibox detector[C]//Computer Vision-ECCV 2016. Amsterdam, The Netherlands: Springer International Publishing, 2016.
[13] LIN T Y, GOYAL P, GIRSHICK R, et al.Focal loss for dense object detection[J].IEEE Transactions on Pattern Analysis & Machine Intelligence, 2017, 42(2):318-327.
[14] GIRSHICK R, DONAHUE J, DARRELL T, et al.Region-based convolutional networks for accurate object detection and segmentation[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2015, 38(1):142-158.
[15] REN S Q, HE K M, GIRSHICK R, et al.Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2016, 39(6): 1137-1149.
[16] HAN K, WANG Y H, TIAN Q, et al.Ghostnet: more features from cheap operations[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Seattle, WA, USA: IEEE, 2020: 1580-1589.
[17] TAN J, YANG Z, REN J, et al.A novel robust low-rank multi-view diversity optimization model with adaptive-weighting based manifold learning[J]. Pattern Recognition, 2022, 122:108298.
[18] CHEN H, DU Y, FU Y, et al.DCAM-Net: A Rapid Detection Network for Strip Steel Surface Defects Based on Deformable Convolution and Attention Mechanism[J]. IEEE Transactions on Instrumentation and Measurement, 2023, 72:1-12.
[19] TIAN R, JIA M.DCC-CenterNet: A Rapid Detection Method for Steel Surface Defects[J]. Measurement, 2022, 187:110211.
[20] DU Y, CHEN H, FU Y, et al.AFF-Net: A Strip Steel Surface Defect Detection Network via Adaptive Focusing Features[J]. IEEE Transactions on Instrumentation and Measurement, 2024, 73:1-14.
[21] QIAN X, WANG., YANG S, et al. LFF-YOLO: A YOLO Algorithm with Lightweight Feature Fusion Network for Multi-Scale Defect Detection[J]. IEEE Access, 2022, 10: 130339-130349.
[22] HU X, LIN S.DFFNet: A Lightweight Approach for Efficient Feature-Optimized Fusion in Steel Strip Surface Defect Detection[J]. Complex & Intelligent Systems, 2024, 10(5): 6705-6723.
[23] HOU Q, ZHOU D, FENG J.Coordinate attention for efficient mobile network design[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2021:13713-13722.
[24] ZHENG Z H, WANG P, LIU W, et al.Distance-IoU loss: faster and better learning for bounding box regression[C]//Proceedings of the AAAI Conference on Artificial Intelligence. 2020, 34(7):12993-13000.
[25] GEVORGYAN Z. SIoU loss: more powerful learning for bounding box regression[R]. arXiv:2205.12740,2022.
[26] AMARI S I.Backpropagation and stochastic gradient descent method[J]. Neurocomputing, 1993, 5(4/5):185-196.
[27] MISRA D.Mish: a self regularized non-monotonic activation function[R]. arXiv:1908.08681, 2019.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

基于深度学习的真空镀膜生产缺陷检测方法研究

The Surface Defect Detection Method of Vacuum Coating Production Based on Deep Learning

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 7

Metrics

本文评价

推荐阅读 10

[1]	唐政. 阵列蒸发源对平移基板成膜均匀性的研究^*[J]. 真空, 2024, 61(6): 21-25.
[2]	李翔, 姜小蛟, 战春鸣, 刘昂, 孙宁, 李加平. 基于系统回归模型的真空镀膜设备液冷加热盘设计方法研究^*[J]. 真空, 2024, 61(4): 6-11.
[3]	李灿伦, 倪俊, 郭腾, 韩玮, 王飞, 祁松松, 李辉, 范孝鹏, 范秋林. 无色聚酰亚胺薄膜二次表面镜的光学特性研究[J]. 真空, 2024, 61(3): 70-73.
[4]	孙彬, 刘兴龙, 徐乘远, 王庆, 蔺增. 真空镀膜助力低碳制造与可持续发展^*[J]. 真空, 2023, 60(3): 12-17.
[5]	邓文宇, 王朋阳, 齐丽君, 段永利, 孙宝玉, 万亿, 张昕洁, 谢元华，杜广煜，刘坤. 钕铁硼永磁材料腐蚀机理及防护研究进展^*[J]. 真空, 2020, 57(5): 45-51.
[6]	王福贞. “热处理技术”和 “真空镀膜技术”在走向融合[J]. 真空, 2020, 57(5): 1-6.
[7]	张粉利, 邓敬莲, 王杰峰, 孟庆远. 钢管镀膜前处理清洗工艺的研究[J]. 真空, 2018, 55(6): 60-63.