基于改进DeepLabV3+网络的光伏组件热斑故障识别及状态量化评估方法研究

陈雷; 刘波; 孙凯; 赵健

doi:10.19912/j.0254-0096.tynxb.2023-1963

PDF(2441 KB)

太阳能学报 ›› 2025, Vol. 46 ›› Issue (3) : 445-453. DOI: 10.19912/j.0254-0096.tynxb.2023-1963

基于改进DeepLabV3+网络的光伏组件热斑故障识别及状态量化评估方法研究

陈雷¹, 刘波¹, 孙凯², 赵健¹

作者信息 +

RESEARCH ON HOT SPOT FAULT IDENTIFICATION METHOD AND STATE QUANTITATIVE EVALUATION OF PHOTOVOLTAIC MODULE BASED ON IMPROVED DEEPLABV3+ NETWORK

Chen Lei¹, Liu Bo¹, Sun Kai², Zhao Jian¹

Author information +

文章历史 +

摘要

针对光伏组件热斑的精确定位和量化评估,提出一种基于改进DeepLabV3+网络与热斑像素比重模型相融合的光伏组件状态量化评估方法,旨在实现不同热斑状态的量化评估。首先,基于获取的红外热斑图像集,提出在DeepLabV3+主干网络中引入迁移学习网络（EfficientNetB7）来提高热斑形状特征提取能力,进而实现热斑的像素级语义分割;其次,利用Canny算法对分割的热斑图像进行像素级轮廓界定,并利用格林积分计算其像素比重;最后,通过构建状态评估模型实现对光伏组件热斑状态的量化评估。现场试验表明,与常见的语义分割方法（DeepLabV3、FCN、U-net、Linknet、SegNet）相比,该文所提方法在像素准确率和平均交并比方面分别达到98.33%和91.43%,具有较好的热斑分割效果。此外,所提状态评估方法可实现对光伏组件热斑大小的准确量化评估。

Abstract

Currently, for the precise location and quantitative assessment of hot spots of PV modules, a method for quantitative assessment of PV module states based on the fusion of improved DeepLabV3+ network and hot spot pixel specific gravity model is proposed, aiming to realize the quantitative assessment of different hot spot states. Firstly, based on the acquired infrared hot spot image set, this paper proposes to introduce a migration learning network （EfficientNetB7） into the DeepLabV3+ backbone network to improve the feature extraction capability, and then realize the pixel-level semantic segmentation of hot spots. Secondly, the Canny algorithm is utilized to define the pixel-level contours of the segmented hot spot images and calculate their pixel weights using Green’s integral. Finally, a state evaluation model is constructed to quantitatively evaluate the hot spot state of PV modules. Field experiments show that compared with common semantic segmentation methods （DeepLabV3, FCN, U-net, Linknet, SegNet）, the proposed method achieves 98.33% pixel accuracy and 91.43% average intersection and merger ratio, which is good for hot spot segmentation. In addition, the proposed state assessment method can realize accurate quantitative assessment of hot spot size of PV modules.

导出引用

陈雷, 刘波, 孙凯, 赵健. 基于改进DeepLabV3+网络的光伏组件热斑故障识别及状态量化评估方法研究[J]. 太阳能学报. 2025, 46(3): 445-453 https://doi.org/10.19912/j.0254-0096.tynxb.2023-1963

Chen Lei, Liu Bo, Sun Kai, Zhao Jian. RESEARCH ON HOT SPOT FAULT IDENTIFICATION METHOD AND STATE QUANTITATIVE EVALUATION OF PHOTOVOLTAIC MODULE BASED ON IMPROVED DEEPLABV3+ NETWORK[J]. Acta Energiae Solaris Sinica. 2025, 46(3): 445-453 https://doi.org/10.19912/j.0254-0096.tynxb.2023-1963

中图分类号： TM615

参考文献

[1] 任惠, 夏静, 卢锦玲, 等. 基于红外图像和改进MobileNet-V3的光伏组件故障诊断方法[J]. 太阳能学报, 2023, 44(8): 238-245.
REN H, XIA J, LU J L, et al.Research on fault diagnosis of photovoltaic modules based on infrared images and improved MobileNet-V3[J]. Acta energiae solaris sinica, 2023, 44(8): 238-245.
[2] 马铭遥, 徐君, 张志祥. 光伏发电系统智能化故障诊断技术[M]. 北京: 机械工业出版社, 2022: 67-85.
MA M Y, XU J, ZHANG Z X.Intelligent fault diagnosis technology for photovoltaic power generation system[M]. Beijing: China Machine Press, 2022: 67-85.
[3] 王乐, 陈雪, 张舒, 等. 光伏组件热斑效应研究[J]. 太阳能学报, 2023, 44(7): 155-161.
WANG L, CHEN X, ZHANG S, et al.Hot spot effect for photovoltaic modules[J]. Acta energiae solaris sinica, 2023, 44(7): 155-161.
[4] 孙海蓉, 周映杰, 张镇韬, 等. 基于改进自私羊群算法的光伏红外热图像热斑识别方法[J]. 中国电机工程学报, 2022, 42(24): 8942-8951.
SUN H R, ZHOU Y J, ZHANG Z T, et al.Hot spot recognition method of photovoltaic infrared thermal image based on improved selfish herd algorithm[J]. Proceedings of the CSEE, 2022, 42(24): 8942-8951.
[5] 赵鑫. 基于GA-AO混合优化算法与BiLSTM的光伏故障诊断[D]. 阜新: 辽宁工程技术大学, 2022.
ZHAO X.Photovoltaic fault diagnosis based on GA-AO hybrid optimization algorithm and BiLSTM[D]. Fuxin: Liaoning Technical University, 2022.
[6] LIU H B, WU Y F, XU X Q, et al.Impulsive neural network-based hot spot fault prediction in photovoltaic power plants[J]. Chinese Journal of power sources, 2022, 46(6): 680-683.
[7] 陈明昊, 孙毅, 谢志远. 基于双层深度强化学习的园区综合能源系统多时间尺度优化管理[J]. 电工技术学报, 2023, 38(7): 1864-1881.
CHEN M H, SUN Y, XIE Z Y.The multi-time-scale management optimization method for park integrated energy system based on the bi-layer deep reinforcement learning[J]. Transactions of China Electrotechnical Society, 2023, 38(7): 1864-1881.
[8] 赵利军, 曹聪颖, 张晋京, 等. 压缩图像增强方法研究综述[J]. 计算机工程与应用, 2023, 59(21): 26-38.
ZHAO L J, CAO C Y, ZHANG J J, et al.Survey of research on compressed image enhancement methods[J]. Computer engineering and applications, 2023, 59(21): 26-38.
[9] 蒋琳, 苏建徽, 李欣, 等. 基于可见光和红外热图像融合的光伏阵列热斑检测方法[J]. 太阳能学报, 2022, 43(1): 393-397.
JIANG L, SU J H, LI X, et al.Hot spot detection of photovoltaic array based on fusion of visible and infrared thermal images[J]. Acta energiae solaris sinica, 2022, 43(1): 393-397.
[10] 刘宇宸, 李浩. 显著性特征融合的热红外图像光伏组件热斑检测[J]. 水力发电, 2023, 49(4): 96-101, 112.
LIU Y C, LI H.Thermal infrared image photovoltaic hot spot detection based on saliency feature fusion[J]. Water power, 2023, 49(4): 96-101, 112.
[11] DI TOMMASO A, BETTI A, FONTANELLI G, et al.A multi-stage model based on YOLOv3 for defect detection in PV panels based on IR and visible imaging by unmanned aerial vehicle[J]. Renewable energy, 2022, 193: 941-962.
[12] 陈博源, 吴佳慧. 基于计算机DCGAN数据增强的YOLOv5光伏电池板热斑检测方法[J]. 软件, 2022, 43(10): 88-91, 101.
CHEN B Y, WU J H.Hot spot detection method of YOLOv5 photovoltaic panel based on computer DCGAN data enhancement[J]. Software, 2022, 43(10): 88-91, 101.
[13] 徐硕, 邓艾东, 杨宏强, 等. 基于改进残差网络的旋转机械故障诊断[J]. 太阳能学报, 2023, 44(7): 409-418.
XU S, DENG A D, YANG H Q, et al.Rotating machinery fault diagnosis method based on improved residual neural network[J]. Acta energiae solaris sinica, 2023, 44(7): 409-418.
[14] 林正文, 宋思瑜, 范钧玮, 等. 改进YOLOv5的光伏组件热斑及遮挡小目标检测[J]. 计算机工程与应用, 2024, 60(1): 84-95.
LIN Z W, SONG S Y, FAN J W, et al.Improved YOLOv5 photovoltaic module thermal spot and occlusion small target detection[J]. Computer engineering and applications, 2024, 60(1): 84-95.
[15] 樊涛, 孙涛, 刘虎. 基于注意力机制的光伏组件热斑检测算法[J]. 北京航空航天大学学报, 2022, 48(7): 1304-1313.
FAN T, SUN T, LIU H.Hot spot detection algorithm of photovoltaic module based on attention mechanism[J]. Journal of Beijing University of Aeronautics and Astronautics, 2022, 48(7): 1304-1313.
[16] 姜萍, 李梦瑶, 栾艳军. 基于改进InceptionV3网络的光伏组件航拍红外图像故障分类方法[J]. 激光杂志, 2022, 43(8): 90-94.
JIANG P, LI M Y, LUAN Y J.Fault classification method of photovoltaic module aerial infrared image based on improved InceptionV3 network[J]. Laser journal, 2022, 43(8): 90-94.
[17] 贾帅男. 基于残差网络的光伏红外图像热斑识别方法[J]. 唐山师范学院学报, 2021, 43(6): 53-56.
JIA S N.Thermal spot recognition of photovoltaic infrared image based on residual network[J]. Journal of Tangshan Normal University, 2021, 43(6): 53-56.
[18] 贾帅康. 基于残差注意力机制的光伏组件热斑图像检测方法研究[D]. 保定: 华北电力大学, 2021.
JIA S K.Research on hot spot image detection method of photovoltaic module based on residual attention mechanism[D]. Baoding: North China Electric Power University, 2021.
[19] 陈剑, 杜文娟, 王海风. 采用深度迁移学习定位含直驱风机次同步振荡源机组的方法[J]. 电工技术学报, 2021, 36(1): 179-190.
CHEN J, DU W J, WANG H F.A method of locating the power system subsynchronous oscillation source unit with grid-connected PMSG using deep transfer learning[J]. Transactions of China Electrotechnical Society, 2021, 36(1): 179-190.
[20] 王艳, 申宗旺, 赵洪山, 等. 基于改进CNN-SVM的光伏组件红外图像故障诊断方法[J]. 华北电力大学学报(自然科学版), 2024, 51(3): 110-117.
WANG Y, SHEN Z W, ZHAO H S, et al.Fault diagnosis method of photovoltaic modules infrared image based on improved CNN-SVM[J]. Journal of North China Electric Power University (natural science edition), 2024, 51(3): 110-117.
[21] 孙海蓉, 潘子杰, 晏勇. 基于深度卷积自编码网络的小样本光伏热斑识别与定位[J]. 华北电力大学学报(自然科学版), 2021, 48(4): 91-98.
SUN H R, PAN Z J, YAN Y.Identification and location of photovoltaic hot spots with small samples based on deep convolutional autoencoder network[J]. Journal of North China Electric Power University (natural science edition), 2021, 48(4): 91-98.
[22] TAN M, LE Q.Efficientnet: rethinking model scaling for convolutional neural networks[C]. International conference on machine learning. PMLR, 2019: 6105-6114.
[23] CHEN L C, ZHU Y K, PAPANDREOU G, et al.Encoder-decoder with atrous separable convolution for semantic image segmentation[M]. Lecture Notes in Computer Science. Cham: Springer International Publishing, 2018: 833-851.
[24] CANNY J. A computational approach to edge detection[J]. IEEE transactions on pattern analysis and machine intelligence, 1986, PAMI-8(6): 679-698.
[25] 谭媛, 黄辉先, 徐建闽, 等. 基于改进Sobel算子的遥感图像道路边缘检测方法[J]. 国土资源遥感, 2016, 28(3): 7-11.
TAN Y, HUANG H X, XU J M, et al.Road edge detection from remote sensing image based on improved Sobel operator[J]. Remote sensing for land & resources, 2016, 28(3): 7-11.
[26] CHEN L C, PAPANDREOU G, SCHROFF F, et al. Rethinking atrous convolution for semantic image segmentation[J]. arXiv e-prints, 2017: arXiv: 1706.05587.
[27] CAI J X, MU T J, LAI Y K, et al.LinkNet: 2D-3D linked multi-modal network for online semantic segmentation of RGB-D videos[J]. Computers & graphics, 2021, 98: 37-47.
[28] RONNEBERGER O, FISCHER P, BROX T.U-net: convolutional networks for biomedical image segmentation[M]. Lecture Notes in Computer Science. Cham: Springer International Publishing, 2015: 234-241.
[29] LONG J, SHELHAMER E, DARRELL T.Fully convolutional networks for semantic segmentation[C]//2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Boston, MA, USA, 2015: 3431-3440.
[30] BHATTASALI T. SEGNET: secure geo-sensor network model [EB/OL].2012:1209.6262. https://arxiv.org/abs/1209.6262v1.
[31] 赵波, 廖坤, 邓春宇, 等. 基于卷积神经学习的光伏板积灰状态识别与分析[J]. 中国电机工程学报, 2019, 39(23): 6981-6989.
ZHAO B, LIAO K, DENG C Y, et al.Image convolutional neural learning based image recognition and analysis method for dust on photovoltaic panel[J]. Proceedings of the CSEE, 2019, 39(23): 6981-6989.
[32] ZHAO J, XU M X, WANG X Y, et al.Data-driven based low-voltage distribution system transformer-customer relationship identification[J]. IEEE transactions on power delivery, 2022, 37(4): 2966-2977.
[33] JIANG Y, ZHAO C H.Attention classification-and-segmentation network for micro-crack anomaly detection of photovoltaic module cells[J]. Solar energy, 2022, 238: 291-304.
[34] VAN QUYEN T, KIM M Y.Feature pyramid network with multi-scale prediction fusion for real-time semantic segmentation[J]. Neurocomputing, 2023, 519: 104-113.