基于CBAM-EfficientNet-B0的铁谱图像磨损类型识别算法
A Wear Type Recognition Algorithm for Ferrography Images Based on CBAM-EfficientNet-B0
DOI: 10.12677/JISP.2022.113012, PDF,   
作者: 刘胜慧：天津工业大学，控制科学与工程学院，天津
关键词: 铁谱分析；磨损类型识别；EfficientNet；深度学习；Ferrographic Analysis； Wear Type Identification； EfficientNet； Deep Learning
摘要: 铁谱图像磨损类型识别是分析机械设备磨损故障的重要方法。针对磨粒数据集样本少且不同磨损类型在质地、形状和颜色上的差异微小导致的分类精度低的问题，提出一种基于改进EfficientNet网络的铁谱图像磨损类型识别算法。本文选择EfficientNet-B0作为磨损类型识别的基本模型，将CBAM注意力模块融合到EfficientNet-B0中，构建CBAM-EfficientNet-B0，从而提高磨粒的聚焦能力和信息表达能力。本文构建了五种磨损类型的磨粒图像数据集。在测试数据集上测试了CBAM-EfficientNet-B0的磨损类型识别能力。实验结果表明，本文提出的磨损类型识别算法CBAM-EfficientNet-B0准确率为92.55%，较改进前EfficientNet-B0算法的准确率提升了2.51%，提高了机械设备磨损状态识别精度和效率。将CBAM-EfficientNet-B0与MobilenetV3、Resnet50、VGG16和ViT分类模型进行比较，实验结果表明CBAM-EfficientNet-B0的精确率、召回率和准确率均高于对比实验中的其他方法。该研究为设备的状态维修和故障诊断提供了新的技术选择。
Abstract: Ferrographic image wear type identification is an important method to analyze the wear failure of mechanical equipment. Aiming at the problem of low classification accuracy caused by the small number of samples in the abrasive particle dataset and the small differences in texture, shape and color of different wear types, a wear type recognition algorithm based on improved EfficientNet network was proposed. In this paper, EfficientNet-B0 is selected as the basic model for wear type recognition, and the CBAM attention module is integrated into EfficientNet-B0 to construct CBAM- EfficientNet-B0, thereby improving the focusing ability and information expression ability of abrasive particles. In this paper, a dataset of abrasive grain images for five types of wear is constructed. The wear type recognition ability of CBAM-EfficientNet-B0 is tested on the test dataset. The experimental results show that the accuracy of the wear type identification algorithm CBAM- EfficientNet-B0 proposed in this paper is 92.55%, which is 2.51% higher than that of the EfficientNet-B0 algorithm before the improvement, which improves the accuracy and efficiency of mechanical equipment wear state identification. Comparing CBAM-EfficientNet-B0 with MobilenetV3, Resnet50, VGG16 and ViT classification models, the experimental results show that the precision, recall and accuracy of CBAM-EfficientNet-B0 are higher than other methods in the comparative experiments. This research provides new technical options for condition maintenance and fault diagnosis of equipment.
文章引用：刘胜慧. 基于CBAM-EfficientNet-B0的铁谱图像磨损类型识别算法[J]. 图像与信号处理, 2022, 11(3): 101-112. https://doi.org/10.12677/JISP.2022.113012

参考文献

[1] 关浩坚, 贺石中, 李秋秋, 杨智宏, 覃楚东, 何伟楚. 卷积神经网络在装备磨损颗粒识别中的研究综述[J]. 摩擦学学报, 2022, 42(2): 426-445.
[2] Wang, S., Wu, T., Wang, K. and Sarkodie-Gyan, T. (2020) Ferrograph Analysis with Improved Particle Segmentation and Classification Methods. Journal of Computing and Information Science in Engineering, 20, Article ID: 021001.
[Google Scholar] [CrossRef] 
[3] Jardine, K., Lin, D. and Banjevic, D. (2006) A Review on Machinery Diagnostics and Prognostics Implementing Condition-Based Maintenance. Mechanical Systems and Signal Processing, 20, 1483-1510.
[Google Scholar] [CrossRef] 
[4] Li, Q., Zhao, T., Zhang, L., Sun, W. and Zhao, X. (2017) Ferrography Wear Particles Image Recognition Based on Extreme Learning Machine. Journal of Electrical and Computer Engineering, 2017, Article ID: 3451358.
[Google Scholar] [CrossRef] 
[5] Peng, P. and Wang, J. (2019) FECNN: A Promising Model for Wear Particle Recognition. Wear, 432-433, Article ID: 202968
[Google Scholar] [CrossRef] 
[6] Jia, F., Wei, H., Sun, H., Song, L. and Yu, F. (2022) An Object Detection Network for Wear Debris Recognition in Ferrography Images. Journal of the Brazilian Society of Mechanical Sciences and Engineering, 44, Article No. 67.
[Google Scholar] [CrossRef] 
[7] 向进. 改进分水岭和灰靶理论在铁谱图像分析中的研究[D]: [硕士学位论文]. 南京: 南京航空航天大学, 2016.
[8] Tan, M. and Le, Q. (2019) Efficientnet: Rethinking Model Scaling for Convolutional Neural Networks. Proceedings of the 36th International Conference on Machine Learning, Vol. 97, Long Beach, 9-15 June 2019, 6105-6114.
[9] Woo, S., Park, J., Lee, J.Y. and Kweon, I.S. (2018) Cbam: Convolutional Block Attention Module. Proceedings of the European Conference on Computer Vision, Munich, 8-14 September 2018, 3-19.
[Google Scholar] [CrossRef] 
[10] Lever, J., Krzywinski, M. and Altman, N. (2016) Classification Evaluation: It Is Important to Understand Both What a Classification Metric Expresses and What It Hides. Nature Methods, 13, 603-605.
[Google Scholar] [CrossRef] 
[11] Liu, W., Anguelov, D., Erhan, D., Szegedy, C., Reed, S., Fu, C.-Y., et al. (2016) SSD: Single Shot MultiBox Detector. European Conference on Computer Vision, Amsterdam, 11-14 October 2016, 21-37.
[Google Scholar] [CrossRef] 
[12] Goyal, P., Dollár, P., Girshick, R., Noordhuis, P., Wesolowski, L., Kyrola, A., et al. (2017) Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour. arXiv preprint arXiv: 1706.02677.
[13] Howard, A., Sandler, M., Chu, G., Wang, W., Chen, L.-C., Tan, M., et al. (2019) Searching for MobileNetV3. Proceedings of the IEEE/CVF International Conference on Computer Vision, Seoul, 27 October-2 November 2019, 1314-1324.
[Google Scholar] [CrossRef] 
[14] Mukti, I.Z. and Biswas, D. (2019) Transfer Learning Based Plant Diseases Detection Using ResNet50. 2019 4th International Conference on Electrical Information and Communication Technology (EICT), Khulna, 20-22 December 2019, 1-6.
[Google Scholar] [CrossRef] 
[15] Qassim, H., Verma, A. and Feinzimer, D. (2018) Compressed Residual-VGG16 CNN Model for Big Data Places Image Recognition. 2018 IEEE 8th Annual Computing and Communication Workshop and Conference (CCWC), Las Vegas, 8-10 January 2018, 169-175.
[Google Scholar] [CrossRef] 
[16] Kolesnikov, A., Dosovitskiy, A., Weissenborn, D., Weissenborn, D., Zhai, X., Unterthiner, T., et al. (2021) An Image Is Worth 16 × 16 Words: Transformers for Image Recognition at Scale. arXiv:2010.11929.