Defect Detection Method for Hydropower Station Equipment Based on Infrared Temperature Measurement Data
-
摘要:
针对目前水电站设备缺陷检测精度低、效率慢、花费时间长和人力成本大等问题，提出了一种基于注意力机制的红外测温数据的水电站设备缺陷检测算法。首先在YOLO（You Only Look Once）v5的特征提取网络中引入新的注意力模块CA（Coordinate Attention），以便更好地捕捉输入中的关键信息，提高模型性能和准确度。其次使用Ghost Convolution和Bottleneck Transformer替换YOLOv5中的普通卷积和C3模块，减少模型的参数量和计算量。最后，将缺陷检测算法和PyQt5界面相结合，实现可视化检测。实验结果表明，改进后的YOLOv5模型的mAP@0.5提升了6.3%，P和R分别提升了6.6%和5.9%，满足水电站设备缺陷的检测需求且具有更高的检测精度。
Abstract:To address the challenges of low accuracy, low efficiency, and significant time and labor costs of the current defect detection process for hydropower station equipment, a defect detection algorithm for hydropower station equipment based on the attention mechanism of infrared temperature measurement data is proposed. First, a new attention module, Coordinate Attention (CA), is introduced into the YOLOv5s feature extraction network to better capture key information in the input and improve the model performance and accuracy. Second, we replaced the ordinary convolution and C3 modules in YOLOv5 with Ghost Revolution and Bottleneck Transformer to reduce the model's parameters and computational complexity. Finally, the defect detection algorithm is combined with the PyQt5 interface to achieve visual image detection and output detection results. The experimental results show that the improved YOLOv5 model's mAP@0.5 is increased by 6.3%, and P and R are increased by 6.6% and 5.9%, respectively, meeting the detection needs of hydropower station equipment defects and achieving a higher detection accuracy.
-
Keywords:
- infrared temperature measurement /
- deep learning /
- YOLOv5 /
- defect detection
-
-
表 1 不同算法的性能比较
Table 1 Performance comparison of different models
Methods mAP/% P/% R/% FPS/(f/s) YOLOv3 21.8 16.3 19.5 35 TPH-YOLOv5 19.2 18.2 24.4 43 Faster-RCNN 12.54 20.12 22.6 32 YOLOv5 26.3 32.8 38.3 38 Ours 32.6 39.4 44.2 45 
下载: 导出CSV
表 2 该数据集消融实验结果
Table 2 Results of ablation experiment of the dataset
Methods mAP@.0.5/% P/% R/% CA Ghost Convolution Bottleneck Transformer √ 28.3 34.6 39.8 √ 27.6 35.4 42.1 √ 27.3 36.5 43.5 √ √ 30.1 37.4 43.6 √ √ √ 32.6 39.4 44.2
下载: 导出CSV
-
[1] 王可为. 红外测温技术在变电运行中研究[J]. 产业创新研究, 2023(14): 138-140. WANG K W. Research on infrared temperature measurement technology in substation operation[J]. Industrial Innovation Research, 2023(14): 138-140.
[2] 邵宇鹰, 王枭, 彭鹏, 等. 基于声成像技术的电力设备缺陷检测方法研究[J]. 中国测试, 2021, 47(7): 42-48. SHAO Y Y, WANG X, PENG P, et al. Research on defect detection method of power equipment based on acoustic imaging technology[J]. China Test, 2021, 47(7): 42-48.
[3] 黄勤琴. 基于红外图像的电力设备及其温度缺陷检测[D]. 成都: 西南交通大学, 2022. HUANG Q Q. Power Equipment and Its Temperature Defect Detection Based on Infrared Image [D]. Chengdu: Southwest Jiaotong University, 2022.
[4] CHENG Y, XIA L, YAN B, et al. A defect detection method based on faster RCNN for power equipment[C]//Journal of Physics: Conference Series. IOP Publishing, 2021, 1754(1): 012025.
[5] WANG K, REN F, KANG X, et al. Thermal defect detection and location for power equipment based on improved VGG16[C]//IEEE International Conference on Agents (ICA), 2021: 41-46.
[6] CHEN X, AN Z, HUANG L, et al. Surface defect detection of electric power equipment in substation based on improved YOLOv4 algorithm[C]//10th International Conference on Power and Energy Systems (ICPES), 2020: 256-261.
[7] WANG C Y, Bochkovskiy A, Liao H Y M. YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023: 7464-7475.
[8] Nooralishahi P, Rezayiye R K, López F, et al. PHM-IRNET: Self-training thermal segmentation approach for thermographic inspection of industrial components[J]. NDT & E International, 2023, 138: 102884.
[9] RUAN L, GAO B, WU S, et al. DeftectNet: Joint loss structured deep adversarial network for thermography defect detecting system[J]. Neurocomputing, 2020, 417: 441-457. DOI: 10.1016/j.neucom.2020.07.093
[10] WANG H, HOU Y, HE Y, et al. A physical-constrained decomposition method of infrared thermography: pseudo restored heat flux approach based on ensemble bayesian variance tensor fraction[J]. IEEE Transactions on Industrial Informatics, 2023, 20(3): 3413-3424.
[11] YU J, HE Y, LIU H, et al. An improved U-Net model for infrared image segmentation of wind turbine blade[J]. IEEE Sensors Journal, 2022, 23(2): 1318-1327.
[12] 余松森, 张明威, 杨欢. 改进YOLOv5的瓷砖表面缺陷检测[J]. 计算机系统应用, 2023, 32(8): 151–161. YU S S, ZHANG M W, YANG H, et al. Improved tile surface defect detection with YOLOv5[J]. Computer System Application, 2023, 32(8): 151-161.
[13] 代牮, 赵旭, 李连鹏, 等. 基于改进YOLOv5的复杂背景红外弱小目标检测算法[J]. 红外技术, 2022, 44(5): 504-512. http://hwjs.nvir.cn/article/id/f71aa5f4-92b0-4570-9056-c2abd5506021 DAI J, ZHAO X, LI L P, et al. Improved YOLOv5-based infrared dim-small target detection under complex background[J]. Infrared Technology, 2022, 44(5): 504-512. http://hwjs.nvir.cn/article/id/f71aa5f4-92b0-4570-9056-c2abd5506021
[14] 王成龙. 基于改进YOLOv5n的绝缘子缺陷检测算法研究[D]. 大庆: 东北石油大学, 2023. WANG C L. Research on Insulator Defect Detection Algorithm Based on Improved YOLOv5n [D]. Daqing: Northeast Petroleum University, 2023.
[15] YAO J, QI J, ZHANG J, et al. A real-time detection algorithm for Kiwifruit defects based on YOLOv5[J]. Electronics, 2021, 10(14): 1711. DOI: 10.3390/electronics10141711
[16] 张玉彬, 刘鹏谦, 陈丽娜, 等. 基于YOLO v5的带涂层钢结构亚表面缺陷脉冲涡流热成像智能检测[J]. 红外技术, 2023, 45(10): 1029-1037. http://hwjs.nvir.cn/article/id/8f53b36b-125e-465a-9f0a-7ea4e570ed41 ZHANG Y B, LIU P Q, CHEN L N, et al. Intelligent detection of subsurface defects of coated Steel structures by pulsed eddy thermal imaging based on YOLO v5[J]. Infrared Technology, 2023, 45(10): 1029-1037. http://hwjs.nvir.cn/article/id/8f53b36b-125e-465a-9f0a-7ea4e570ed41
[17] Srinivas A, Lin T Y, Parmar N, et al. Bottleneck transformers for visual recognition[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021: 16519-16529.
[18] 叶舒然, 张珍, 王一伟, 等. 基于卷积神经网络的深度学习流场特征识别及应用进展[J]. 航空学报, 2021, 42(4): 185-199. YE S R, ZHANG Z, WANG Y W, et al. Deep learning flow field feature recognition and application progress based on convolutional neural network [J]. Acta Aeronautica et Astronautica, 2021, 42(4): 185-199.
[19] Willman J. Modern PyQt[M]. Berkeley: Apress, 2021: 1-42.
[20] 王永平, 张红民, 彭闯, 等. 基于YOLOv3的高压开关设备异常发热点目标检测方法[J]. 红外技术, 2020, 42(10): 983-987. http://hwjs.nvir.cn/article/id/hwjs202010011 WANG Y P, ZHANG H M, PENG C, et al. The target detection method for abnormal heating point of high-voltage switchgear based on YOLOv3 [J]. Infrared Technology, 2020, 42(10): 983-987. http://hwjs.nvir.cn/article/id/hwjs202010011
-
期刊类型引用(1)
1. 张锐舟，张丕状，姚金杰，陈明，刘畅. 创新光电二极管前置放大电路设计方法. 国外电子测量技术. 2019(11): 71-76 . 
百度学术
其他类型引用(3)
计量
- 文章访问数: 31
- HTML全文浏览量: 11
- PDF下载量: 8
- 被引次数: 4
