Scholarlink

The estimation of series resistance in photovoltaic (PV) cells is a vital parameter that substantially influences their efficiency and performance. This work presents a novel method to predict the slope of a Current-Voltage (I-V) curve of a PV cell in the first quadrant. The value of this slope is directly related to the series resistance experienced by the PV cell. Utilizing artificial intelligence, we have developed a Convolutional model capable of estimating this slope using electroluminescence (EL) images of the cells. The model was trained on sample images of PV cells featuring artificial defects, along with the corresponding slope values computed from the I-V curves of the cells. The presented model has demonstrated low error values across various metrics (MAE of 5.041), showcasing its accuracy and reliability. Furthermore, it has been compared with other machine learning methods, highlighting its competitive performance. This approach provides a promising tool for improving the assessment and diagnosis of PV cell efficiency and reliability, potentially leading to enhanced performance and longevity of photovoltaic systems.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Novel Convolutional Neural Network Model for Estimating Series Resistance in PV Cells by Predicting I-V Curve Slopes on Electroluminescence Images

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Hector Felipe Mateo-Romero,
  • Jose Ignacio Morales Aragonés,
  • Luis Hernández-Callejo,
  • Miguel Ángel González-Rebollo,
  • Valentín Cardeñoso-Payo,
  • Victor Alonso-Gómez,
  • Emmanuel Ahatsi,
  • Ginés García Mateos

摘要

The estimation of series resistance in photovoltaic (PV) cells is a vital parameter that substantially influences their efficiency and performance. This work presents a novel method to predict the slope of a Current-Voltage (I-V) curve of a PV cell in the first quadrant. The value of this slope is directly related to the series resistance experienced by the PV cell. Utilizing artificial intelligence, we have developed a Convolutional model capable of estimating this slope using electroluminescence (EL) images of the cells. The model was trained on sample images of PV cells featuring artificial defects, along with the corresponding slope values computed from the I-V curves of the cells. The presented model has demonstrated low error values across various metrics (MAE of 5.041), showcasing its accuracy and reliability. Furthermore, it has been compared with other machine learning methods, highlighting its competitive performance. This approach provides a promising tool for improving the assessment and diagnosis of PV cell efficiency and reliability, potentially leading to enhanced performance and longevity of photovoltaic systems.