光伏组件在运行过程中可能出现“热斑效应”,为及时检测热斑,提出一种基于SSD改进的热斑缺陷检测方法。将SSD骨干神经网络更换为ResNet 101,并提出一种新型注意力网络,该网络能学习同一通道特征图区域间关系。实验表明：改进后的SSD 300模型average precision 50（AP50）由96.3%提升至97.7%。针对热斑缺陷红外图像稀少的问题,提出一种新型图像增强方法,扩充热斑红外图像数据集,改进SSD 300模型AP50进一步提升至97.9%,且检测速率仍达到23.6 FPS,该方法基本满足实际应用需求。
Abstract
Hot spot problem may occur during the operation of photovoltaic panels. In order to detect hotspot, an improved hotspot detection method based on Single Shot Multibox Detector（SSD） is proposed. SSD's backbone is replaced by ResNet 101, and a new attention network is proposed that can learn the relationships among regions of the same channel feature map. The experimental results show that the improved SSD 300 model's average precision 50 （AP50） increases from 96.3% to 97.7%. To conquer the shortage of hotspot infrared images, a new method of data augmentation is proposed. After using that method, the improved SSD 300's AP50 furtherly increases to 97.9%, and detection speed reaches 23.6 FPS, the improved model basically meets the requirements of practical application.
关键词
光伏组件 /
深度学习 /
目标检测 /
红外图像 /
SSD /
热斑
Key words
PV modules /
deep learning /
object detection /
infrared images /
SSD /
hotspot
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 刘恒, 张志祥, 马铭遥, 等. 基于I-V特性分析的晶硅光伏组件故障诊断[J]. 太阳能学报, 2021, 42(6): 130-137.
LIU H, ZHANG Z X, MA M Y, et al.Fault diagnosis of crystalline silicon photovoltaic module based on I-V characteristic analysis[J]. Acta energiae solaris sinica, 2021, 42(6): 130-137.
[2] 云平, 刘恒, 张志祥. 基于I-V曲线的光伏组件热斑测试与分析[J]. 太阳能, 2019(10): 33-39.
YUN P, LIU H, ZHANG Z X.Test and analysis of hot spot on PV modules based on I-V curve[J]. Solar energy, 2019(10): 33-39.
[3] 蒋琳, 苏建徽, 施永, 等. 基于红外热图像处理的光伏阵列热斑检测方法[J]. 太阳能学报, 2020, 41(8): 180-184.
JIANG L, SU J H, SHI Y, et al.Hot spots detection of operating PV arrays through IR thermal image[J]. Acta energiae solaris sinica, 2020, 41(8): 180-184.
[4] 王阿勇, 杨静. 太阳能组件热斑检测中的动态模板匹配方法研究[J]. 西安理工大学学报, 2012, 28(4): 474-477.
WANG A Y, YANG J.Research on the dynamic template matching method in solar energy component hot spot detection[J]. Journal of Xi'an University of Technology, 2012, 28(4): 474-477.
[5] 位硕权. 基于红外图像的光伏组件热斑智能检测[D]. 杭州: 浙江大学, 2020.
WEI S Q.Intelligent detection of hotspots of photovoltaic modules based on infrared images[D]. Hangzhou: Zhejiang University, 2020.
[6] 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.
[7] 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.
[8] LIU W, AMGUELOV D, ERHAN D, et al.SSD: single shot multibox detector[C]//Computer Vision and Pattern Recognition 2015, Boston, MA, USA, 2015.
[9] LAW H, DENG J.CornerNet: detecting objects as paired keypoints[C]//Computer Vision and Pattern Recognition 2018, Salt Lake City, UT, USA, 2018.
[10] SIMONYAN K, ZISSERMAN A.Very deep convolutional networks for large-scale image recognition[C]//International Conference on Learning Representations(ICLR), 2015: 1-14.
[11] HE K M, ZHANG X Y, REN S Q, et al.Deep residual learning for image recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, USA, 2016: 770-778.
[12] 郭梦浩, 徐红伟. 基于Faster RCNN 的红外热图像热斑缺陷检测研究[J]. 计算机系统应用, 2019, 28(11): 265-270.
GUO M H, XU H W.Hot spot defect detection based on infrared thermal image and Faster RCNN[J]. Computer systems & applications, 2019, 28(11): 265-270.
[13] 王道累, 李超, 李明山, 等. 基于深度卷积神经网络的光伏组件热斑检测[J]. 太阳能学报, 2022, 43(1): 412-417.
WANG D L, LI C, LI M S, et al.Solar photovoltaic modules hot spot detection based on deep convolutional neural networks[J]. Acta energiae solaris sinica, 2022, 43(1): 412-417.
[14] HU J, SHEN L, SUN G.Squeeze-and-excitation networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Salt Lake City, UT, USA, 2018: 7132-7141.
[15] ZHANG S F, ZHU X Y, LEI Z, et al.S3FD: single shot scale-invariant face detector[C]//International Conference on Computer Vision, Venice, Italy, 2017: 192-201.
[16] FU C Y, LIU W, RANGA A, et al.DSSD:deconvolutional single shot detector[EB/OL]. https://arxiv.org/abs/1701.06659.
基金
国家自然科学基金（12172210; 61502297）
PDF(2053 KB)
