PvIR-DCANet Based Infrared Image Defect Detection for Photovoltaic Modules

LYU Xiaojie; PAN Wei; WANG Fan; CHAI Shaobo; TIAN Yizhi

doi:10.20204/j.sp.2025.08013

English Version Reader LoginJournal Login

您当前的位置：

首页 >

文章列表页 >

PvIR-DCANet Based Infrared Image Defect Detection for Photovoltaic Modules

更新时间：2025-10-24

- PvIR-DCANet Based Infrared Image Defect Detection for Photovoltaic Modules
- Vol. 53, Issue 8, Pages: 105-113(2025)
- 作者机构：
  
  1. 广东电网有限责任公司广州供电局,广东,广州,510620
  2. 天津天大求实电力新技术股份有限公司,天津,300380
  3. 新疆大学电气工程学院,新疆,乌鲁木齐,830000
- 作者简介：
- 基金信息：
- DOI：10.20204/j.sp.2025.08013
  CLC：
- Published：2025
- 稿件说明：
移动端阅览
吕晓杰,潘威,汪帆,柴绍博,田易之.基于PvIR-DCANet的光伏组件红外图像缺陷检测[J].智慧电力,2025,53(8):105-113.

LYU Xiaojie, PAN Wei, WANG Fan, et al. PvIR-DCANet Based Infrared Image Defect Detection for Photovoltaic Modules[J]. 2025, 53(8): 105-113.
吕晓杰,潘威,汪帆,柴绍博,田易之.基于PvIR-DCANet的光伏组件红外图像缺陷检测[J].智慧电力,2025,53(8):105-113. DOI： 10.20204/j.sp.2025.08013.

LYU Xiaojie, PAN Wei, WANG Fan, et al. PvIR-DCANet Based Infrared Image Defect Detection for Photovoltaic Modules[J]. 2025, 53(8): 105-113. DOI： 10.20204/j.sp.2025.08013.

摘要

针对可见光图像在光伏组件缺陷检测中因光照不均与反射干扰导致精度下降的问题，提出一种基于坐标注意力与双向特征融合的光伏红外图像缺陷检测网络PvIR-DCANet。该网络在骨干网络中引入坐标注意力机制（CAM），增强对细微缺陷的检测能力；设计双向密集特征金字塔网络（BDFPN），强化多尺度特征融合以提升不同尺度缺陷的检测性能；同时采用SIoU损失函数优化密集小尺度缺陷目标的定位精度。实验结果表明，PvIR-DCANet的mAP达92.0%，检测速度为57.9 FPS，在精度与实时性之间实现了良好平衡。该方法兼具高准确性与工程实用性，适用于复杂场景下的光伏组件高效缺陷检测任务。

Abstract

To address the accuracy degradation caused by uneven illumination and reflection interference in visible-light-based defect detection of photovoltaic （PV） modules， this study proposes PvIR-DCANet—an infrared image defect detection network incorporating coordinated attention and bidirectional feature fusion. The network introduces a coordinated attention mechanism （CAM） into the backbone to enhance sensitivity to subtle defects. A bidirectional dense feature pyramid network （BDFPN） is designed to strengthen multi-scale feature fusion， improving detection performance across varying defect sizes. Simultaneously， the SIoU loss function optimizes localization accuracy for densely distributed small-scale defects. Experimental results demonstrate that PvIR-DCANet achieves 92.0% mAP with 57.9 FPS detection speed， striking an optimal balance between precision and real-time performance. This method exhibits high accuracy and engineering practicality for efficient PV module defect detection in complex scenarios.

关键词

Keywords

references

王琳，侯涛，王得鹏，等. 基于4G通信的接触网绝缘子在线监测装置研制[J]. 电网与清洁能源，2023，39（4）: 31-36，46.

刘树林，庹汉宇，张元昌. 二次侧附加电容磁复位正激变换器电气性能分析及参数设计[J]. 电工技术学报，2024，39（19）: 6008-6017.

彭昱，黄守道，何赟泽，等. 电机绕组缺陷的电-磁-热多物理场仿真与定量检测方法[J]. 电机与控制学报，2021，25（1）: 79-89.

PRAMANIK A，PAL S K，MAITI J，et al. Granulated RCNN and multi-class deep sort for multi-object detection and tracking[J]. IEEE Transactions on Emerging Topics in Computational Intelligence，2021，6（1）: 171-181.

王辉，欧阳缮，刘庆华，等. 基于深度学习的探地雷达二维剖面图像结构特征检测方法[J]. 电子与信息学报，2022，44（4）: 1284-1294.

张文彪，马永华，白晓静，等. 基于改进Faster R-CNN的太阳能电池板缺陷识别[J]. 电网技术，2022，46（7）: 2593-2600.

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.

张林鍹，郑兴，陈飞，等. 深度学习的红外热成像电路板元器件识别研究[J]. 电子科技大学学报，2024，53（4）: 560-567.

汤璐，王淑青，王年涛，等. 基于改进YOLOX网络的雾天绝缘子缺陷检测[J]. 高压电器，2024，60（3）: 223-228.

金鑫，程凌森，赵亮，等. 基于改进YOLOv5模型的智能变电站目标违规行为检测方法研究[J]. 电测与仪表，2024，61（8）: 179-185.

WANG Y，SHEN L，LI M，et al. PV-YOLO: lightweight YOLO for photovoltaic panel fault detection[J]. IEEE Access，2023，11: 10966-10976.

杨长春，贺轩轩，王睿，等. 基于改进YOLOv8光伏板缺陷检测算法[J]. 电子测量技术，2024，47（23）: 181-192.

李利荣，陈鹏，张云良，等. 基于改进CenterNet的输电线路电力器件及异常目标检测[J]. 高电压技术，2023，49（11）: 4757-4765.

艾上美，周剑峰，张必朝，等. 基于改进SSD算法的光伏组件缺陷检测研究[J]. 智慧电力，2023，51（12）: 53-58.

SU B，CHEN H，LIU K，et al. RCAG-Net: Residual channelwise attention gate network for hot spot defect detection of photovoltaic farms[J]. IEEE Transactions on Instrumentation and Measurement，2021，70: 1-14.

YANG S，AN W，LI S，et al. An improved FCOS method for ship detection in SAR images[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing，2022，15: 8910-8927.

YADAV A，KUMAR E. Object detection on real-time video with FPN and modified Mask RCNN based on inception-ResNetV2[J]. Wireless Personal Communications，2024，138（4）: 2065-2090.

AI J，QU Z，ZHAO Z，et al. An SAR target classification algorithm based on the central coordinate attention module[J]. IEEE Sensors Journal，2023，24（2）: 1941-1952.

HOUDED K V，KAMBLE P M，HEGADI R S. A comparative analysis of iou based losses with YOLOv8: fruit tree detection from uav images[J]. Procedia Computer Science，2025，260: 898-905.

BHUYAN P，SINGH P K，DAS S K. Res4net-CBAM: A deep cnn with convolution block attention module for tea leaf disease diagnosis[J]. Multimedia Tools and Applications，2024，83（16）: 48925-48947.

宋智伟，黄新波，纪超，等. 基于Flexible YOLOv7的输电线路绝缘子缺陷检测和故障预警方法[J]. 高电压技术，2023，49（12）: 5084-5094.

ZHU X，GUO K，REN S，et al. Lightweight image super-resolution with expectation-maximization attention mechanism[J]. IEEE Transactions on Circuits and Systems for Video Technology，2021，32（3）: 1273-1284.

SAPKOTA R，FLORES-CALERO M，QURESHI R，et al. YOLO advances to its genesis: a decadal and comprehensive review of the You Only Look Once （YOLO） series[J]. Artificial Intelligence Review，2025，58（9）: 1-83.

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Short-term Power Load Forecasting Based on VMD-TCN-BiLSTM-Attention

Photovoltaic Power Forecasting Based on Deep Perception-Interaction Transformer

Energy Disaggregation of Commercial Buildings Based on Attention-LSTM

Non-invasive Household Load Decomposition Based on GRNN and Attention Mechanism Model

Image Segmentation Method of Transmission Line Insulator Based on Improved U-Net

Related Author

DAI Jie

LI Guoliang

LIU Yiyan

YANG Zhenzhen

ZHANG Yi

CAO Zhiyuan

SHI Yan

ZHANG Luxi

Related Institution

College of Energy and Electrical Engineering，Chang’an University

国网内蒙古东部电力有限公司供电服务监管与支持中心，内蒙古通辽 0280002.南京邮电大学理学院

State Grid Shanghai Electrical Power Research Institute

School of Electrical Engineering， Shanghai University of Electric Power

（College of Electrical Engineering，Guizhou University，，）

AI问答

⁰