Fault Detection and Identification of Multi-Source Insulators Based on Improved YOLOv7
-
摘要:
为提高复杂背景下异源绝缘子的故障检测准确率，本文提出一种基于异源图像下的改进YOLOv7模型的绝缘子故障识别方法。为突出绝缘子的位置以及故障信息对异源绝缘子图像进行配准融合，为降低计算复杂度以及获得更高的可移植性，将原YOLOv7的主干特征提取网络换为MOBELINET网络，为减少复杂背景下绝缘子的漏检、误检等问题，将原YOLOv7的损失函数由Complete-intersection-Over-Union（CIOU）改为FOICAL-EIOU进一步提高模型预测框的回归效果。最后在YOLOv7检测头部分引入可变形卷积Deformable Convolution Network2（DCNv2）加强对不同尺度大小绝缘子发热故障区域的适应能力。实验结果表明改进的模型Mean Average Precision（mAP）值为96.6%，比原YOLOv7模型mAP值提高9.9%，参数量下降了30.5%，浮点运算数下降了49.2%，较YOLOV5、YOLOV8目标检测模型mAP值分别提高12.2%、12.4%。所提出的改进模型可以有效实现异源绝缘子的故障检测与识别。
Abstract:To improve the accuracy of fault detection for heterogeneous insulators with complex backgrounds, an improved YOLOv7 model-based insulator fault recognition method based on heterogeneous images is proposed in this paper. An image of a heterogeneous insulator was registered and fused to highlight the location of the insulators and fault information. Then, to reduce the computational complexity and improve portability, the original backbone feature extraction network of YOLOv7 was replaced by the MOBELINET network. To reduce the problems of missing and false detections of insulators in complex backgrounds, the original loss function of YOLOv7 was changed from CIOU to FOICAL-EIOU to further improve the regression effect of the model prediction box. Finally, a deformable convolutional DCNv2 was introduced in the YOLOv7 detection head to enhance its adaptability to insulator heating fault areas of different scales. The experimental results show that the mAP value of the improved model is 96.6%, which is 9.9% higher than that of the original YOLOv7 model, the number of parameters decreases by 30.5%, and the floating-point operator decreases by 49.2%, which are 12.2% and 12.4% higher than those of the YOLOV5 and YOLOV8 target detection models, respectively. The proposed improved model could effectively detect and identify heterogeneous insulators.
-
Keywords:
- insulator /
- deformable convolution /
- loss function /
- YOLOv7 /
- fault detection
-
-
表 1 图像数据组成
Table 1 Composition of image data
Infrared image 2640 Visible image 7110（Contains an IR-paired 2640 Visible image） Normal image Interior overheated image Normal image Exterior broken image 1584 1056 4977 2133 
下载: 导出CSV
表 2 实验参数
Table 2 Experimental parameters
Experimental values The number of values Epoch 150 Batch size 16 Learning rate 0.001 Optimizer SGD Size/pixel 640
下载: 导出CSV
表 3 消融实验结果
Table 3 Results of ablation experiment
Test MOBELINET FOICAL-EIOU DCNv2 Parameter/M GFLOPs @mAP0.5 1 6.02 13.2 86.7% 2 √ 4.17 7.0 85.5% 3 √ 6.15 12.1 97.8% 4 √ 6.02 12.0 86.0% 5 √ √ √ 4.18 6.7 96.6%
下载: 导出CSV
表 4 不同检测模型对比结果
Table 4 Comparison results of different detection models
Name Parameter/M GFLOPs Precision Recall SSD 25.6 33.6 61.2% 61.0% YOLOV5 7.06 16.5 84.6% 91.9% YOLOv7 6.02 13.2 85.2% 91.1% YOLOV8 3.00 8.1 79.8% 84.2% Ours 4.18 6.7 92.1% 94.9%
下载: 导出CSV
-
[1] 周自强, 陈强, 马必焕, 等. 一种改进的YOLO目标检测方法在电缆设备异常状态识别中的应用[J]. 电测与仪表, 2020, 57(2): 14-20. ZHOU Ziqiang, CHEN Qiang, MA Bihuan, et al. An improved YOLO target detection method with its application in cable device abnormal condition recognition[J]. Electrical Measurement & Instrumentation, 2020, 57(2): 14-20.
[2] 姜文志, 李炳臻, 顾佼佼, 等. 基于改进YOLO V3的舰船目标检测算法[J]. 电光与控制, 2021, 28(6): 52-56, 67. DOI: 10.3969/j.issn.1671-637X.2021.06.012 JIANG Wenzhi, LI Bingzhen, GU Jiaojiao, et al. A ship target detection algorithm based on improved YOLOv3[J]. Electronics Optics & Control, 2021, 28(6): 52-56, 67. DOI: 10.3969/j.issn.1671-637X.2021.06.012
[3] 王凯, 王伟, 蒋志伟. 基于改进YOLO模型的遥感小目标检测[J]. 电光与控制, 2023, 30(6): 60-64. DOI: 10.3969/j.issn.1671-637X.2023.06.010 WANG Kai, WANG Wei, JIANG Zhiwei. Remote sensing small target detection based on improved YOLO model[J]. Electronics Optics & Control, 2023, 30(6): 60-64. DOI: 10.3969/j.issn.1671-637X.2023.06.010
[4] 毛建旭, 贺振宇, 王耀南, 等. 电力巡检机器人路径规划技术及应用综述[J]. 控制与决策, 2023, 38(11): 3009-3024. MAO Jianxu, HE Zhenyu, WANG Yaonan, et al. Review of research and applications on path planning technology for power inspection robots[J]. Control and Decision, 2023, 38(11): 3009-3024. .
[5] 王义军, 曹培培, 王雪松, 等. 基于深度学习的绝缘子自爆检测方法研究[J]. 东北电力大学学报, 2020, 40(3): 33-40. WANG Yijun, CAO Peipei, WANG Xuesong, et al. Research on insulator self explosion detection method based on deep learning[J]. Journal of Northeast Dianli University, 2020, 40(3): 33-40.
[6] 邱刚, 陈杰, 张廼龙, 等. 基于YOLOv3和HSV颜色空间的绝缘子低/零值故障诊断研究[J]. 高压电器, 2023, 59(1): 148-153. QIU Gang, CHEN Jie, ZHANG Nailong, et al. Research of low/zero defect diagnosis of porcelain insulator based on YOLOv3 and HSV color space image processing technology[J]. High Voltage Apparatus, 2023, 59(1): 148-153.
[7] 陈俊杰, 叶东华, 产焰萍, 等. 基于Faster R-CNN模型的绝缘子故障检测[J]. 电工电气, 2020(4): 56-60, 73. CHEN Junjie, YE Donghua, CHAN Yanping, et al. Insulator fault detection based on faster R-CNN[J]. Electrotechnics Electric, 2020(4): 56-60, 73.
[8] HU S, ZHAO F, LU H, et al. Improving YOLOv7-tiny for infrared and visible light image object detection on drones[J]. Remote Sensing, 2023, 15(13): 3214. DOI: 10.3390/rs15133214
[9] 吕志宁. 输电线路常见故障分析与检测方法综述[J]. 自动化与仪器仪表, 2020(1): 161-164, 8. LYU Zhining. A survey of common faults analysis and detection methods for transmission line[J]. Automation & Instrumentation, 2020(1): 161-164, 8.
[10] 李云红, 刘宇栋, 苏雪平, 等. 红外与可见光图像配准技术研究综述[J]. 红外技术, 2022, 44(7): 641-651. http://hwjs.nvir.cn/cn/article/id/77ef812e-5018-435f-a023-771b550bedc7 LI Yunhong, LIU Yudong, SU Xueping, et al. Review of infrared and visible image registration[J]. Infrared Technology, 2022, 44(7): 641-651. http://hwjs.nvir.cn/cn/article/id/77ef812e-5018-435f-a023-771b550bedc7
[11] 徐永会, 杨德智, 刘芳名. 基于对数极坐标和频域率的互信息图像配准[J]. 舰船电子工程, 2022, 42(11): 86-89, 150. XU Yonghui, YANG Dezhi, LIU Fangming. Mutual information image registration based on logarithmic polar coordinates and frequency domain rate[J]. Ship Electronic Engineering, 2022, 42(11): 86-89, 150.
[12] WANG R, LIANG F, MOU X, et al. Development of an improved YOLOv7-based model for detecting defects on strip steel surfaces[J]. Coatings, 2023, 13(3): 536. DOI: 10.3390/coatings13030536
[13] 陈学深, 吴昌鹏, 党佩娜, 等. 基于改进轻量化YOLOv4模型的虾只肉壳辨识方法[J]. 农业工程学报, 2023, 39(9): 278-286. CHEN Xueshen, WU Changpeng, DANG Peina, et al. Improved lightweight YOLOv4 model-based method for the identification of shrimp flesh and shell[J]. Transactions of the Chinese Society of Agricultural Engineering, 2023, 39(9): 278-286.
[14] Howard A, Sandler M, Chu G, et al. Searching for mobilenetv3[C]// Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019: 1314-1324.
[15] 许爱华, 陈佳韵, 张明文, 等. 基于改进的YOLO v4绝缘子目标识别算法研究[J]. 吉林大学学报(信息科学版), 2023, 41(3): 545-551. DOI: 10.3969/j.issn.1671-5896.2023.03.020 XU Aihua, CHEN Jiayun, ZHANG Mingwen, et al. Research on insulator detection algorithm based on improved YOLO v4[J]. Journal of Jilin University (Information Science Edition), 2023, 41(3): 545-551. DOI: 10.3969/j.issn.1671-5896.2023.03.020
[16] 刘宇, 雷雪梅. 融合MobileNetV3特征的结构化剪枝方法[J]. 上海交通大学学报, 2023, 57(9): 1203-1213. LIU Yu, LEI Xuemei. A structured pruning method integrating characteristics of MobileNetV3[J]. Journal of Shanghai Jiaotong University, 2023, 57(9): 1203-1213.
[17] 赵文清, 刘亮, 胡嘉伟, 等. 基于深度可分离空洞卷积金字塔的变压器渗漏油检测[J]. 智能系统学报, 2023, 18(5): 966-974. ZHAO Wenqing, LIU Liang, HU Jiawei, et al. Detection of transformer oil leakage based on deep separable atrous convolution pyramid[J]. CAAI Transactions on Intelligent Systems, 2023, 18(5): 966-974.
[18] 刘雄彪, 杨贤昭, 陈洋, 等. 基于CIoU改进边界框损失函数的目标检测方法[J]. 液晶与显示, 2023, 38(5): 656-65. LIU Xiongbiao, YANG Xianzhao, CHEN Yang, et al. Object detection method based on CIoU improved bounding box loss function[J]. Chinese Journal of Liquid Crystals and Displays, 2023, 38(5): 656-65.
[19] 周敏. 基于Focal-EIOU函数的被动式太赫兹图像违禁物品识别[J]. 太赫兹科学与电子信息学报, 2022, 20(8): 810-816. ZHOU Min. Detection of prohibited objects in passive terahertz images based on Focal-EIOU loss function[J]. Journal of Terahertz Science and Electronic Information Technology, 2022, 20(8): 810-816.
[20] ZHANG Y F, REN W, ZHANG Z, et al. Focal and efficient IOU loss for accurate bounding box regression[J]. Neurocomputing, 2022, 506: 146-157.
[21] 粟兴旺, 王晓明, 黄金玻, 等. 基于可变形卷积与注意力机制的X光安检违禁品检测[J]. 电子测量技术, 2023, 46(10): 98-108. SU Xingwang, WANG Xiaoming, HUANG Jinbo, et al. Prohibited items detection based on deformable convolution and attention mechanism in X-ray security inspection[J]. Electronic Measurement Technology, 2023, 46(10): 98-108.
[22] ZHU X, HU H, LIN S, et al. Deformable convnets v2: More deformable, better results[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019: 9308-9316.
[23] 曹伊宁, 李超, 彭雅坤. 基于改进YOLOv7的夜间行人检测算法[J]. 长江信息通信, 2022, 35(10): 57-60. CAO Yining, LI Chao, PENG Yakun. Night pedestrian detection algorithm based on improved YOLOv7[J]. Changjiang Information & Communications, 2022, 35(10): 57-60.
计量
- 文章访问数: 39
- HTML全文浏览量: 5
- PDF下载量: 8
