DETECT PV CELL DEFECTION BASED ON ELECTROLUMINESCENCE LIGHT USING DEEP LEARNING

Nguyễn Quốc Minh; Lê Thị Minh Châu; Nguyễn Đăng Tiến; Lê Minh Hiếu

doi:10.34238/tnu-jst.4511

DETECT PV CELL DEFECTION BASED ON ELECTROLUMINESCENCE LIGHT USING DEEP LEARNING

About this article

Received: 17/05/21 Revised: 15/07/21 Published: 21/07/21

Authors

1. Nguyen Quoc Minh , Hanoi University of Science & Technology
2. Le Thi Minh Chau, Hanoi University of Science & Technology
3. Nguyen Dang Tien, Hanoi University of Science & Technology
4. Le Minh Hieu, Hanoi University of Science & Technology

Abstract

PV cell surface defects inspection is essential to maintain safety, reliability and optimal efficiency of the solar plant. Cell surface defection is the most popular type of fault and it can be hardly detected by normal visual inspection. Defected cells without detection and maintainane can affect the performance of other normal cells since they are connected in series and parallel in large number. In this research, we present a method to automatically detect PV cell surface defection using image processing technique by deep learning model. The input data include 2146 high resolution electroluminescence images of mono and poly PV cells. This type of image is usually captured by drones. The images are then fed into convolutional neural network (CNN) for feature extraction and support vector machine (SVM) for image classification. The results show that the proposed deep learning model can classify the normal and surface defect cells with the accuracy of 91.63%.

Keywords

PV cell; Electroluminescence light; Convolutional neural network; Support vector machine; Deep learning

Full Text:

PDF (Tiếng Việt)

References

[1] Q. K. Nguyen, Annual Report on Electric Power Generation Development Scenerios in Vietnam, Green Innovation and Delelopment Centre of Vietnam, 2019.

[2] M. Köntges, S. Kurtz, C. Packard, U. Jahn, K. Berger, K. Kato, T. Friesen, H. Liu, and M. Iseghem, “Review of Failures of Photovoltaic Modules,” IEA-PVPS Report, International Energy Agency, Paris, France, 2014.

[3] J. S. Fada, M. A. Hossain, J. L. Braid, S. Yang, T. J. Peshek, and R. H. French, "Electroluminescent Image Processing and Cell Degradation Type Classification via Computer Vision and Statistical Learning Methodologies," 2017 IEEE 44th Photovoltaic Specialist Conference (PVSC), Washington, DC, USA, 2017, pp. 3456-3461.

[4] A. Bartler, L. Mauch, B. Yang, M. Reuter, and L. Stoicescu, "Automated Detection of Solar Cell Defects with Deep Learning," 2018 26th European Signal Processing Conference (EUSIPCO), Rome, Italy, 2018, pp. 2035-2039.

[5] S. Deitsch et al., “Automatic classification of defective photovoltaic module cells in electroluminescence images,” Solar Energy, vol. 185, pp. 455-468, 2019.

[6] V. S. B. Kurukuru, A. Haque, M. A. Khan, and A. K. Tripathy, "Fault classification for Photovoltaic Modules Using Thermography and Machine Learning Techniques," International Conference on Computer and Information Sciences (ICCIS), Saudi Arabia, 2019, pp. 1-6.

[7] M. Tan and Q. V. Le, “EfficientNet: Rethinking model scaling for convolutional neural networks,” International Conference on Machine Learning, 2019.

[8] S. Deitsch, Buerhop-Lutz, and C., Sovetkin, “Segmentation of photovoltaic module cells in uncalibrated electroluminescence images,” Machine Vision and Applications, vol. 84, pp. 1-23, 2021.

[9] A. Géron, Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems, 2^nd edition, O'Reilly Media, ISBN: 9781492032649, 2019.

DOI: https://doi.org/10.34238/tnu-jst.4511

Refbacks

There are currently no refbacks.



Remember me