Empirical Analysis of AI-Based Hotspot Detection in Photovoltaic Panels Using Thermal Images

K. Subha, S. Sharanya

2025

Abstract

The abundant heat energy exhaled from the sun can be used for diverse applications using modern technology, which is known as solar energy. Photoelectric cell converts sunlight energy into electrical energy directly by using a photovoltaic effect. Long-term exposure of PV panels to malicious conditions increases their risk of cell damage which causes hot-spot and can reduce the efficiency of electricity production, perhaps resulting in fires. Surface fault detection (FD) is a key strategy to enhance PV panel reliability and performance and improve energy management. This study presents an empirical analysis of hotspot detection in PV panels using thermal images through different machine learning (ML) and deep learning (DL) algorithms including Vision Transformer (ViT) that were assessed for fault detection in solar cells. A dataset consisting of thermal images, derived from the solar plant, was utilized in this study, consisting of 3 classes: cell, hot spot and multi hot spot for experiment fault classification in solar panels. To evaluate performance a comprehensive comparison of accuracy, precision, f1-score, recall, mAP (Mean Average Precision) parameters of the model were used. The result showed that the model based on the vision transformer exhibited better performance in hotspot fault detection problems in PV modules. In fact, transformer models were found to be efficient for fault detection with good accuracy (98%). Through empirical analysis it was found that Transformer based techniques have outperformed well based on ML, DL-based approaches.

Paper Citation

in Harvard Style

Subha K. and Sharanya S. (2025). Empirical Analysis of AI-Based Hotspot Detection in Photovoltaic Panels Using Thermal Images. In Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25; ISBN 978-989-758-777-1, SciTePress, pages 400-407. DOI: 10.5220/0013930500004919

in Bibtex Style

@conference{icrdicct`2525,
author={K. Subha and S. Sharanya},
title={Empirical Analysis of AI-Based Hotspot Detection in Photovoltaic Panels Using Thermal Images},
booktitle={Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25},
year={2025},
pages={400-407},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013930500004919},
isbn={978-989-758-777-1},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25
TI - Empirical Analysis of AI-Based Hotspot Detection in Photovoltaic Panels Using Thermal Images
SN - 978-989-758-777-1
AU - Subha K.
AU - Sharanya S.
PY - 2025
SP - 400
EP - 407
DO - 10.5220/0013930500004919
PB - SciTePress