PHOTOVOLTAIC POWER LOSS PREDICTION METHOD BASED ON FUSION ATTENTION MECHANISM AND IMPROVED YOLOv8 MULTIMODAL MODEL

Wang Daolei; Fang Mengxin; Zhang Zhen; Zhu Rui; Zhao Wenbin

doi:10.19912/j.0254-0096.tynxb.2024-0892

PDF(1834 KB)

Acta Energiae Solaris Sinica ›› 2025, Vol. 46 ›› Issue (9) : 602-610. DOI: 10.19912/j.0254-0096.tynxb.2024-0892

PHOTOVOLTAIC POWER LOSS PREDICTION METHOD BASED ON FUSION ATTENTION MECHANISM AND IMPROVED YOLOv8 MULTIMODAL MODEL

Wang Daolei, Fang Mengxin, Zhang Zhen, Zhu Rui, Zhao Wenbin

Author information +

History +

Abstract

In order to solve the problem of power loss of photovoltaic modules caused by the“hot spot effect”, a multi-modal YOLOv8n-cls model prediction method based on Super Token Attention Mechanism （STA） was proposed. The STA attention mechanism was added to the image classification network YOLOv8n-cls to solve the problem that the neural network tend to capture the local hot spot features in the shallow layer with high redundancy. At the same time, a multi-modal model of infrared image and digital information fusion of photovoltaic modules was built based on the YOLOv8n-cls algorithm to explore the influence of irradiance parameters on power loss prediction. A photovoltaic multi-source dataset composed of photovoltaic infrared images, power loss and irradiance parameter labels was constructed, and power loss prediction experiments were carried out. Experiments show that the Top1 accuracy of the improved network model for the hot spot effect caused by leaves, dust and guano reaches 95.6%, 89.0% and 88.1% respectively, which is 3.3%, 10.0% and 7.5% higher than that of the original model, which proves the superiority of the proposed algorithm in predicting the power loss of photovoltaic modules with hot spot effect.

Key words

image classification / photovoltaics / solar irradiance / power loss / YOLOv8 / multimodal

Cite this article

EndNote

Ris (Procite)

Bibtex

Download Citations

Wang Daolei, Fang Mengxin, Zhang Zhen, Zhu Rui, Zhao Wenbin. PHOTOVOLTAIC POWER LOSS PREDICTION METHOD BASED ON FUSION ATTENTION MECHANISM AND IMPROVED YOLOv8 MULTIMODAL MODEL[J]. Acta Energiae Solaris Sinica. 2025, 46(9): 602-610 https://doi.org/10.19912/j.0254-0096.tynxb.2024-0892

References

[1] 梁作. “双碳” 目标下光伏产业发展的新机遇与新应用[J]. 广东经济, 2023(17): 74-77.
LIANG Z. New opportunities and new applications for the development of the photovoltaic industry under the goal of "dual carbon"[J]. Guangdong economy, 2023(17): 74-77.
[2] 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.
[3] 文贤馗,何明君,张俊玮,等.基于K均值聚类的光伏集群发电功率超短期预测研究[J].电力系统保护与控制,2025,53(12):165-172.
WEN X K,HE M J,ZHANG J W,et al.Research on ultra-short-term power forecasting of photovoltaic clusters based on K-means clustering[J].Power system protection and Control,2025,53(12):165-172
[4] 吐松江·卡日, 吴现, 马小晶, 等. 基于地基云图数据多维特征融合的光伏功率预测算法[J]. 电力系统保护与控制, 2025, 53(11): 84-94.
TUSONGJIANG K,WU X, MA X J,et al.Photovoltaic power prediction algorithm based on multidimensional features fusion of ground-based cloud images[J]. Power system protection and control,2025,53(11):84-94
[5] MEHTA S, AZAD A P, CHEMMENGATH S A, et al.DeepSolarEye: power loss prediction and weakly supervised soiling localization via fully convolutional networks for solar panels[C]//2018 IEEE Winter Conference on Applications of Computer Vision (WACV),Lake Tahoe, NV, USA, 2018: 333-342.
[6] LONG J, SHELHAMER E, DARRELL T.Fully convolutional networks for semantic segmentation[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Boston, MA, USA, 2015: 3431-3440.
[7] ZHANG W J, LIU S Q, GANDHI O, et al.Deep-learning-based probabilistic estimation of solar PV soiling loss[J]. IEEE transactions on sustainable energy, 2021, 12(4): 2436-2444.
[8] JIAO X, LI X S, YANG Y H, et al.Novel and comprehensive approach for power loss estimation of soiled photovoltaic modules[J]. Solar energy, 2024, 268: 112283.
[9] NGUYEN-DUC T, LE-VIET T, NGUYEN-DANG D, et al.Photovoltaic array reconfiguration under partial shading conditions based on short-circuit current estimated by convolutional neural network[J]. Energies, 2022, 15(17): 6341.
[10] 章涛, 柳玉宾, 崔承刚, 等. 基于语义分割的光伏组件积灰检测与分析[J]. 科学技术与工程, 2022, 22(32): 14259-14266.
ZHANG T, LIU Y B, CUI C G, et al. Detection and analysis of photovoltaic module ash accumulation based on semantic segmentation[J]. Science technology and engineering, 2022, 22(32): 14259-14266.
[11] 伊纪禄, 刘文祥, 马洪斌, 等. 太阳电池热斑现象和成因的分析[J]. 电源技术, 2012, 36(6): 816-818.
YI J L, LIU W X, MA H B, et al. Solar hot spot phenomenon and its analysis[J]. Chinese journal of power sources, 2012, 36(6): 816-818.
[12] 李志刚, 田盛. 局部阴影下光伏阵列结构优化[J]. 太阳能学报, 2016, 37(12): 2999-3004.
LI Z G, TIAN S. Structure optimazation of PV array under partial shade[J]. Acta energiae solaris sinica, 2016, 37(12): 2999-3004.
[13] REDMON J, DIVVALA S, GIRSHICK R, et al.You only look once: unified, real-time object detection[C]//2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA, 2016: 779-788.
[14] HUANG H B, ZHOU X Q, CAO J, et al.Vision transformer with super token sampling[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, Vancouver, BC, Canada, 2023: 22690-22699.
[15] CHEN W, WANG W P, LIU L, et al.New ideas and trends in deep multimodal content understanding: a review[J]. Neurocomputing, 2021, 426: 195-215.
[16] JAMPANI V, SUN D Q, LIU M Y, et al.Superpixel sampling networks[C]//Proceedings of the European Conference on Computer Vision(ECCV), Munich, Germany, 2018: 352-368.
[17] WANG C X, JIA M S, LI M R, et al.Attention is all you need for blind room volume estimation[C]//ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Seoul, Korea, Republic of, 2024: 1341-1345.
[18] LIU Q M, CHMELY S C, ABDOULMOUMINE N.Biomass treatment strategies for thermochemical conversion[J]. Energy & fuels, 2017, 31(4): 3525-3536.
[19] ZLATIĆ L. An alternative for one-hot encoding in neural network models[EB/OL].2023: 2311.05911. https://arxiv.org/abs/2311.05911v1.
[20] WANG B, SHAABAN K, KIM I.Revealing the hidden features in traffic prediction via entity embedding[J]. Personal and ubiquitous computing, 2021, 25(1): 21-31.
[21] 姚宏民, 杜欣慧, 秦文萍. 基于密度峰值聚类及GRNN神经网络的光伏发电功率预测方法[J]. 太阳能学报, 2020, 41(9): 184-190.
YAO H M, DU X H, QIN W P. PV power forecasting approach based on density peaks clustering and general regression neural network[J]. Acta energiae solaris sinica, 2020, 41(9): 184-190.