DUST ACCUMULATION DETECTION ON PHOTOVOLTAIC MODULES BASED ON IMPROVED YOLOv5 MODEL

Li Le; Liu Yifei; Li Haoyi; Wu Xinyue; Yin Xiaoxuan; Wu Haoyue

doi:10.19912/j.0254-0096.tynxb.2025-0807

PDF(10957 KB)

Acta Energiae Solaris Sinica ›› 2026, Vol. 47 ›› Issue (9) : 352-361. DOI: 10.19912/j.0254-0096.tynxb.2025-0807

DUST ACCUMULATION DETECTION ON PHOTOVOLTAIC MODULES BASED ON IMPROVED YOLOv5 MODEL

Li Le¹, Liu Yifei¹, Li Haoyi², Wu Xinyue³, Yin Xiaoxuan¹, Wu Haoyue⁴

Author information +

History +

Abstract

To address the decline in power generation efficiency and the shortened operational lifespan of photovoltaic （PV） modules caused by surface dust accumulation, this paper proposes an improved YOLOv5-based model for dust detection. The proposed model enhances the original YOLOv5 architecture by introducing multi-scale GS convolution modules and VoVGSCSP structures, incorporating the CBAM attention mechanism, and modifying the SPPF module to improve both accuracy and speed. Based on a sufficiently large and representative dataset, CLAHE, sharpening, and Gamma parameter adjustment are collaboratively applied to part of the dataset. An experimental platform was developed in PyCharm to implement and verify the improved model, enabling efficient detection of dust locations and regional dust accumulation on PV module surfaces. Experimental results show that the improved YOLOv5 model achieves 3.3 percentage points increase in m_AP@0.5∶0.95 over the baseline YOLOv5, while the FPS improves from 65.2 FPS to 68.3 FPS, demonstrating advantages in both accuracy and speed. This study develops a surface dust detection model for photovoltaic modules under varying illumination conditions, providing methodological support for operation and maintenance management and cleaning decision-making in photovoltaic power plants.

Key words

PV modules / deep learning / object detection / regional dust accumulation / intelligent detection

Cite this article

EndNote

Ris (Procite)

Bibtex

Download Citations

Li Le, Liu Yifei, Li Haoyi, Wu Xinyue, Yin Xiaoxuan, Wu Haoyue. DUST ACCUMULATION DETECTION ON PHOTOVOLTAIC MODULES BASED ON IMPROVED YOLOv5 MODEL[J]. Acta Energiae Solaris Sinica. 2026, 47(9): 352-361 https://doi.org/10.19912/j.0254-0096.tynxb.2025-0807

References

[1] 李浩义, 刘云鹏, 李乐, 等. 基于静电吸附机理的光伏组件无水化除尘方法研究[J]. 太阳能学报, 2025, 46(7): 606-613.
Li H Y, Liu Y P, Li L, et al.Research on anhydration dust removal method for photovoltaic modules based on electrostatic adsorption mechanism[J]. Acta Energiae Solaris Sinica, 2025, 46(7): 606-613.
[2] 闫群民, 屈晨光, 王磊, 等. 考虑双碳目标的新型电力系统源-网-荷-储评价体系研究[J]. 太阳能学报, 2025, 46(12): 469-482.
Yan Q M, Qu C G, Wang L, et al.Study on source-grid-load-storage evaluation system of new-type power system under carbon peaking and carbon-neutral goal[J]. Acta Energiae Solaris Sinica, 2025, 46(12): 469-482.
[3] Li W Y, Dong F G, Shi M X, et al.Multi-attribute decision-making research on investment suitability assessment of hydropower-wind-photovoltaic-storage complementary system based on dynamic social network[J]. Energy Conversion and Management, 2024, 307: 118358.
[4] 李乐, 尹晓萱, 武文华, 等. 适用于光伏板静电除尘的纳米银线-环氧树脂透明导电薄膜研究[J]. 中国电机工程学报, 2024, 44(23): 9463-9474.
Li L, Yin X X, Wu W H, et al.Research on silver nanowire-epoxy resin transparent conductive film for electrostatic dust removal technology of solar photovoltaic panel[J]. Proceedings of the CSEE, 2024, 44(23): 9463-9474.
[5] 曲宏伟, 王靖雯. 积灰对光伏板输出特性影响理论和试验研究[J]. 太阳能学报, 2018, 39(8): 2335-2340.
Qu H W, Wang J W.Theory and experimental research of effect of dust deposition on output characteristics of PV module[J]. Acta Energiae Solaris Sinica, 2018, 39(8): 2335-2340.
[6] 徐志明, 曲宏伟, 王靖雯, 等. 太阳能光伏板积灰的最佳清扫周期[J]. 中国电机工程学报, 2018, 38(6): 1738-1742.
Xu Z M, Qu H W, Wang J W, et al.Dusty PV modules optimal cleaning cycle[J]. Proceedings of the CSEE, 2018, 38(6): 1738-1742.
[7] 殷孝雎, 于金池, 郝志鹏, 等. 基于光伏组件图像特征的故障检测方法[J]. 太阳能学报, 2025, 46(4): 273-279.
Yin X J, Yu J C, Hao Z P, et al.Fault detection method based on image features of photovoltaic modules[J]. Acta Energiae Solaris Sinica, 2025, 46(4): 273-279.
[8] Ghahremani A, Adams S D, Norton M, et al.Advancements in AI-Driven detection and localization of solar panel defects[J]. Advanced Engineering Informatics, 2025, 64: 103104.
[9] 杨瑞珍, 杜博伦, 何赟泽, 等. 晶体硅光伏电池电磁感应激励红外热辐射缺陷检测与成像技术[J]. 电工技术学报, 2018, 33(S2): 321-330.
Yang R Z, Du B L, He Y Z, et al.Infrared radiation defect detection and imaging technique under active electromagnetic induction excitation for crystalline silicon photovoltaic cells[J]. Transactions of China Electrotechnical Society, 2018, 33(S2): 321-330.
[10] 王道累, 姚勇, 张世恒, 等. 基于红外热图像的光伏组件热斑深度学习检测方法[J]. 中国电机工程学报, 2023, 43(24): 9608-9616.
Wang D L, Yao Y, Zhang S H, et al.Deep learning detection method of photovoltaic module hot spot based on infrared thermal image[J]. Proceedings of the CSEE, 2023, 43(24): 9608-9616.
[11] Huang J, Zeng K Y, Zhang Z J, et al.Solar panel defect detection design based on YOLOv5 algorithm[J]. Heliyon, 2023, 9(8): e18826.
[12] Sun T Y, Xing H S, Cao S X, et al.A novel detection method for hot spots of photovoltaic (PV) panels using improved anchors and prediction heads of YOLOv5 network[J]. Energy Reports, 2022, 8: 1219-1229.
[13] Pan W W, Sun X F, Wang Y L, et al.Enhanced photovoltaic panel defect detection via adaptive complementary fusion in YOLO-ACF[J]. Scientific Reports, 2024, 14: 26425.
[14] 陈海永, 袁乐, 王世杰, 等. 基于多尺度编码互补注意力网络的光伏缺陷检测[J]. 太阳能学报, 2023, 44(10): 191-197.
Chen H Y, Yuan L, Wang S J, et al.Photovoltaic defect detection based on multi-scale coding complementary attention network[J]. Acta Energiae Solaris Sinica, 2023, 44(10): 191-197.
[15] 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.
[16] 孙鹏翔, 毕利, 王俊杰. 基于改进深度残差网络的光伏板积灰程度识别[J]. 计算机应用, 2022, 42(12): 3733-3739.
Sun P X, Bi L, Wang J J.Dust accumulation degree recognition of photovoltaic panel based on improved deep residual network[J]. Journal of Computer Applications, 2022, 42(12): 3733-3739.
[17] Yang Y H, Miao Z L, Zhang H, et al.Lightweight attention-guided YOLO with level set layer for landslide detection from optical satellite images[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2024, 17: 3543-3559.
[18] 王育欣, 张志, 张家亮, 等. 基于改进YOLOv5算法的光伏缺陷检测[J]. 太阳能学报, 2024, 45(12): 139-145.
Wang Y X, Zhang Z, Zhang J L, et al.Photovoltaic defect detection based on improved YOLOv5 algorithm[J]. Acta Energiae Solaris Sinica, 2024, 45(12): 139-145.
[19] 项新建, 汤卉, 肖家乐, 等. 基于多尺度特征融合SSDLite的光伏组件缺陷检测[J]. 太阳能学报, 2025, 46(1): 669-675.
Xiang X J, Tang H, Xiao J L, et al.Defect detection of photovoltaic modules based on multi-scale feature fusion SSDLite[J]. Acta Energiae Solaris Sinica, 2025, 46(1): 669-675.
[20] 崔沛然, 崔明建, 汪清, 等. 基于张量分解个性化联邦学习的网络-光伏爬坡协同攻击辨识[J]. 电工技术学报, 2025, 40(23): 7677-7693.
Cui P R, Cui M J, Wang Q, et al.Cyber-ramping coordinated attack identification method for PV using a tensor decomposition based personalized federated learning[J]. Transactions of China Electrotechnical Society, 2025, 40(23): 7677-7693.
[21] 梁远升, 程康, 王钢, 等. 基于概率预测与随机响应面法的新能源孤岛配电网实时风险评估与调控策略[J]. 电网技术, 2023, 47(12): 4948-4957.
Liang Y S, Cheng K, Wang G, et al.Real-time risk assessment and regulation strategy of new energy islanded distribution network based on probabilistic prediction and stochastic response surface methodology[J]. Power System Technology, 2023, 47(12): 4948-4957.
[22] 李毅, 徐慧英, 朱信忠, 等. 基于YOLOv5n模型改进的口罩检测算法: Mask-YOLO[J]. 计算机工程, 2025, 51(6): 297-310.
Li Y, Xu H Y, Zhu X Z, et al.Mask-YOLO: improved mask detection algorithm based on YOLOv5n[J]. Computer Engineering, 2025, 51(6): 297-310.
[23] Jia Y F, Chen G D, Zhao L Q.Defect detection of photovoltaic modules based on improved VarifocalNet[J]. Scientific Reports, 2024, 14: 15170.
[24] 米增, 连哲. 面向通用目标检测的YOLO方法研究综述[J]. 计算机工程与应用, 2024, 60(21): 38-54.
Mi Z, Lian Z.Review of YOLO methods for universal object detection[J]. Computer Engineering and Applications, 2024, 60(21): 38-54.