基于改进YOLOX的变电站仪表外观缺陷检测算法

doi:10.11996/JG.j.2095-302X.2023050937

图学学报 ›› 2023, Vol. 44 ›› Issue (5): 937-946.DOI: 10.11996/JG.j.2095-302X.2023050937

• 图像处理与计算机视觉 • 上一篇下一篇

基于改进YOLOX的变电站仪表外观缺陷检测算法

赵振兵¹^,²^,³(), 马迪雅¹, 石颖¹, 李刚⁴

1.华北电力大学电子与通信工程系，河北保定 071003
2.华北电力大学河北省电力物联网技术重点实验室，河北保定 071003
3.华北电力大学复杂能源系统智能计算教育部工程研究中心，河北保定 071003
4.华北电力大学控制与计算机工程学院，河北保定 071003

收稿日期:2023-03-31 接受日期:2023-06-26 出版日期:2023-10-31 发布日期:2023-10-31
作者简介:赵振兵(1979-)，男，教授，博士。主要研究方向为电力视觉技术等。E-mail：zhaozhenbing@ncepu.edu.cn
基金资助:
国家自然科学基金项目(61871182);国家自然科学基金项目(U21A20486);河北省自然科学基金项目(F2020502009);河北省自然科学基金项目(F2021502008);河北省自然科学基金项目(F2021502013)

Appearance defect detection algorithm of substation instrument based on improved YOLOX

ZHAO Zhen-bing¹^,²^,³(), MA Di-ya¹, SHI Ying¹, Li Gang⁴

1. Department of Electronic and Communication Engineering, North China Electric Power University, Baoding Hebei 071003, China
2. Hebei Key Laboratory of Power Internet of Things Technology, North China Electric Power University, Baoding Hebei 071003, China
3. Engineering Research Center of Intelligent Computing for Complex Energy Systems, Ministry of Education, North China Electric Power University, Baoding Hebei 071003, China
4. School of Control and Computer Engineering, North China Electric Power University, Baoding Hebei 071003, China

Received:2023-03-31 Accepted:2023-06-26 Online:2023-10-31 Published:2023-10-31
About author:ZHAO Zhen-bing (1979-), professor, Ph.D. His main research interests cover computer vision technology in electric power system, etc. E-mail：zhaozhenbing@ncepu.edu.cn
Supported by:
National Natural Science Foundation of China(61871182);National Natural Science Foundation of China(U21A20486);Natural Science Foundation of Hebei Province(F2020502009);Natural Science Foundation of Hebei Province(F2021502008);Natural Science Foundation of Hebei Province(F2021502013)

摘要/Abstract

摘要：

针对变电站仪表外观缺陷检测任务中存在的缺陷分布广、缺陷外观特征变化多样与特征提取困难问题，提出了基于改进YOLOX的变电站仪表外观缺陷检测算法。通过对比样本策略寻找到各类外观缺陷中有判别力的特征并对其进行深入挖掘；接着在网络的骨干部分加入全局上下文信息模块(GC-Model)，增强模型对于外观缺陷特征的学习能力，提高网络性能；最后在预测部分使用SIoU损失函数，充分考虑方向框的角度问题对于模型优化的影响，从而提高对变电站仪表外观缺陷检测的准确率，减少漏检、误检情况发生。在实验中将选取4种典型的变电站仪表外观缺陷类型作为实验对象，通过实验结果分析，相较基线模型，平均精确率(mAP)提升了6.2%，可以有效提升变电站仪表外观缺陷的检测效果，为无人变电站的智能监测提供了有利条件。

关键词: 变电站仪表, 外观缺陷检测, YOLOX, 对比样本策略, 上下文信息模块, SIoU

Abstract:

In response to the wide distribution of defects, the variety of appearance defect features, and the challenges in extracting features during the substation instrument appearance defect detection task, the appearance defect detection algorithm of substation instrument based on improved YOLOX was proposed. Through the contrast sample strategy, the discriminative features of various appearance defects could be identified and extracted in depth. Additionally, the Global Context Feature Module was added to the network backbone to enhance the learning ability of the model for appearance defect features and improve the network performance. Finally, the SIoU loss function was integrated in the prediction part to fully consider the influence of the direction frame’s angle on model optimization, thereby improving the accuracy of detecting appearance defects in substation instrument and reducing the occurrence of missed detection and false detection. In the experiment, four typical types of appearance defects of substation instruments were selected as the experimental objects. The analysis of experimental results showed an average accuracy rate increase of 6.2% compared with the baseline model. This substantial improvement effectively enhanced the detection effect for appearance defects in substation instruments, thus creating favorable conditions for intelligent monitoring of unmanned substations.

Key words: substation instruments, appearance defect detection, YOLOX, contrast sample strategy, global context feature module, SIoU

中图分类号:

TP391

赵振兵, 马迪雅, 石颖, 李刚. 基于改进YOLOX的变电站仪表外观缺陷检测算法[J]. 图学学报, 2023, 44(5): 937-946.

ZHAO Zhen-bing, MA Di-ya, SHI Ying, Li Gang. Appearance defect detection algorithm of substation instrument based on improved YOLOX[J]. Journal of Graphics, 2023, 44(5): 937-946.

图/表 10

图1 缺陷数据集中的部分仪表示例((a)表盘脏污；(b)表盘破损；(c)表盘凝露；(d)示数不清晰)

Fig. 1 Example of some instruments in defect dataset ((a) Dial dirty; (b) Dial obscure; (c) Dial condensation; (d) Indicator confusion)

图2 整体网络架构

Fig. 2 Overall network architecture

图3 全局上下文信息模块整体结构

Fig. 3 Overall structure of global context feature module

表1 实验环境

Table 1 Experimental environment

名称	型号
操作系统	Ubuntu18.04
CPU	I9-10900X
GPU	RTX A5000
CUDA	11.1
cuDNN	7.6.5
Python	3.7
深度学习框架	PyTorch 1.8.1

表2 数据集详细情况

Table 2 Dataset details

缺陷种类	图像数量	标签名称
表盘脏污	100	dirty
表盘破损	70	obscure
表盘凝露	72	condensation
示数不清晰	87	confuse
正常仪表	329	meter

表3 消融实验结果

Table 3 Ablation test results

YOLOX	对比样本策略	全局上下文信息模块	SIoU损失函数	AP (%)				mAP (%)
YOLOX	对比样本策略	全局上下文信息模块	SIoU损失函数	表盘脏污	表盘破损	表盘凝露	示数不清晰	mAP (%)
√	-	-	-	91.2	84.2	84.3	76.7	84.1
√	√	-	-	90.8	83.5	91.3	83.6	87.3
√	-	√	-	91.4	87.9	84.7	84.5	87.1
√	-	-	√	89.8	89.4	85.9	83.3	87.1
√	√	√	√	92.4	87.0	96.3	85.4	90.3

图4 多个模型的精度曲线

Fig. 4 Precision curves for multiple models

图5 多个模型的损失曲线

Fig. 5 Loss curves for multiple models

表4 与先进检测算法性能对比

Table 4 Performance comparison with advanced detect algorithms

方法	AP (%)				mAP (%)	FPS
方法	表盘脏污	表盘破损	表盘凝露	示数不清晰	mAP (%)	FPS
YOLOX	91.2	84.2	84.3	76.7	84.1	91
Faster R-CNN	86.3	89.7	90.9	89.9	89.0	3
YOLOF	76.4	70.7	79.2	80.1	76.4	22
YOLOv6	88.5	73.0	82.1	88.5	79.8	244
本文算法	92.4	87.0	96.3	85.4	90.3	82

图6 不同模型测试结果((a)表盘脏污；(b)表盘破损；(c)表盘凝露；(d)示数不清晰)

Fig. 6 Test results of different models ((a) Dial dirty; (b) Dial obscure; (c) Dial condensation; (d) Indicator confusion)

参考文献 23

[1]	TANG W J, CHEN H G. Research on intelligent substation monitoring by image recognition method[J]. International Journal of Emerging Electric Power Systems, 2021, 22(1): 1-7. DOI URL
[2]	赵书涛, 李宝树, 崔桂彦, 等. 基于计算机视觉的远程变电站状态监测与诊断新策略[J]. 电网技术, 2005, 29(6): 63-66.
	ZHAO S T, LI B S, CUI G Y, et al. Remote state monitoringand diagnosis of substation based on computer vision[J]. Power System Technology, 2005, 29(6): 63-66. (in Chinese)
[3]	李大伟, 贾鹏飞, 李卫国, 等. 一种基于卡尔曼滤波与模糊算法的变电站机器人组合导航及控制系统设计[J]. 智能系统学报, 2013, 8(3): 226-232.
	LI D W, JIA P F, LI W G, et al. A kind of integrated navigation and control system design for substation robot based on the Kalman filtering and fuzzy algorithm[J]. CAAI Transactions on Intelligent Systems, 2013, 8(3): 226-232. (in Chinese)
[4]	马鹏, 樊艳芳. 基于深度迁移学习的小样本智能变电站电力设备部件检测[J]. 电网技术, 2020, 44(3): 1148-1159.
	MA P, FAN Y F. Small sample smart substation power equipment component detection based on deep transfer learning[J]. Power System Technology, 2020, 44(3): 1148-1159. (in Chinese)
[5]	SHU Y, YIN H R, RAJABI M, et al. RNA-based micelles: a novel platform for paclitaxel loading and delivery[J]. Journal of Controlled Release, 2018, 276: 17-29. DOI URL
[6]	毛爱坤, 刘昕明, 陈文壮, 等. 改进YOLOv5算法的变电站仪表目标检测方法[J]. 图学学报, 2023, 44(3): 448-455. DOI
	MAO A K, LIU X M, CHEN W Z, et al. Improved substation instrument target detection method for YOLOv5 algorithm[J]. Journal of Graphics, 2023, 44(3): 448-455. (in Chinese)
[7]	LIU Z, LIN Y T, CAO Y, et al. Swin transformer: hierarchical vision transformer using shifted windows[C]// 2021 IEEE/CVF International Conference on Computer Vision. New York: IEEE Press, 2022: 9992-10002.
[8]	MA N N, ZHANG X Y, ZHENG H T, et al. ShuffleNet V2: practical guidelines for efficient CNN architecture design[EB/OL]. [2023-01-20]. https://arxiv.org/abs/1807.11164.
[9]	范新南, 黄伟盛, 史朋飞, 等. 基于改进YOLOv4的嵌入式变电站仪表检测算法[J]. 图学学报, 2022, 43(3): 396-403.
	FAN X N, HUANG W S, SHI P F, et al. Embedded substation instrument detection algorithm based on improved YOLOv4[J]. Journal of Graphics, 2022, 43(3): 396-403. (in Chinese)
[10]	齐冬莲, 韩译锋, 周自强, 等. 基于视频图像的输变电设备外部缺陷检测技术及其应用现状[J]. 电子与信息学报, 2022, 44(11): 3709-3720.
	QI D L, HAN Y F, ZHOU Z Q, et al. Review of defect detection technology of power equipment based on video images[J]. Journal of Electronics & Information Technology, 2022, 44(11): 3709-3720. (in Chinese)
[11]	CUI Y, HUANG X, ZHANG X, et al. A defects detection system for substation based on YOLOX[C]// The 5th International Electrical and Energy Conference. New York: IEEE Press, 2022: 4703-4707.
[12]	LV S G, LIU K, QIAO Y H, et al. Automatic defect detection based on improved Faster RCNN for substation equipment[J]. Journal of Physics: Conference Series, 2020, 1544(1): 12157. DOI
[13]	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 and Machine Intelligence, 2017, 39(6): 1137-1149. DOI PMID
[14]	YING Y, WANG Y Z, YAN Y F, et al. An improved defect detection method for substation equipment[C]// The 39th Chinese Control Conference. New York: IEEE Press, 2020: 6318-6323.
[15]	WANG J, ZHANG Q W. Visual defect detection for substation equipment based on joint inspection data of camera and robot[C]// The 5th Information Technology and Mechatronics Engineering Conference. New York: IEEE Press, 2020: 491-495.
[16]	CAO Y, XU J R, LIN S, et al. GCNet: non-local networks meet squeeze-excitation networks and beyond[C]// 2019 IEEE/CVF International Conference on Computer Vision Workshop. New York: IEEE Press, 2020: 1971-1980.
[17]	GEVORGYAN Z. SIoU loss: more powerful learning for bounding box regression[EB/OL]. [2023-01-20]. https://arxiv.org/abs/2205.12740.
[18]	CHEN Q, WANG Y M, YANG T, et al. You only look one-level feature[C]// 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2021: 13034-13043.
[19]	LI C Y, LI L L, JIANG H L, et al. YOLOv6: a single-stage object detection framework for industrial applications[EB/OL]. [2023-01-20]. https://arxiv.org/abs/2209.02976.
[20]	GE Z, LIU S T, WANG F, et al. YOLOX: exceeding YOLO series in 2021[EB/OL]. [2023-01-20]. https://arxiv.org/abs/2107.08430.
[21]	JAISWAL A, BABU A R, ZADEH M Z, et al. A survey on contrastive self-supervised learning[EB/OL]. [2023-01-20]. https://arxiv.org/abs/2011.00362.
[22]	LIN T Y, GOYAL P, GIRSHICK R, et al. Focal loss for dense object detection[C]// 2017 IEEE International Conference on Computer Vision. New York: IEEE Press, 2017: 2999-3007.
[23]	罗会兰, 陈鸿坤. 基于深度学习的目标检测研究综述[J]. 电子学报, 2020, 48(6): 1230-1239. DOI
	LUO H L, CHEN H K. Survey of object detection based on deep learning[J]. Acta Electronica Sinica, 2020, 48(6): 1230-1239. (in Chinese)

基于改进YOLOX的变电站仪表外观缺陷检测算法

Appearance defect detection algorithm of substation instrument based on improved YOLOX

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 23

相关文章 6

编辑推荐

Metrics

本文评价

[1]	曹义亲, 周一纬, 徐露. 基于E-YOLOX的实时金属表面缺陷检测算法[J]. 图学学报, 2023, 44(4): 677-690.
[2]	毛爱坤, 刘昕明, 陈文壮, 宋绍楼. 改进YOLOv5算法的变电站仪表目标检测方法[J]. 图学学报, 2023, 44(3): 448-455.
[3]	罗文宇, 傅明月. 基于YoloX-ECA模型的非法野泳野钓现场监测技术[J]. 图学学报, 2023, 44(3): 465-472.
[4]	赵辉 , 赵尧 , 金林林 , 董兰芳 , 肖潇 . 基于 YOLOX 的小目标烟火检测技术研究与实现[J]. 图学学报, 2022, 43(5): 783-790.
[5]	胡海涛 , 杜昊晨 , 王素琴 , 石敏 , 朱登明 , . 改进 YOLOX 的药品泡罩铝箔表面缺陷检测方法[J]. 图学学报, 2022, 43(5): 803-814.
[6]	范新南, 黄伟盛, 史朋飞, 辛元雪, 朱凤婷, 周润康. 基于改进 YOLOv4 的嵌入式变电站仪表检测算法[J]. 图学学报, 2022, 43(3): 396-403.