Joint-Embedding Predictive Architecture for Solar PV Panel Fault Classification

2026-07-10

AI Summary

This work addresses the challenges of class imbalance, limited texture information, and subtle thermal feature differences in fault classification of photovoltaic panels using thermal infrared imagery. To this end, the authors propose JEFFNet, a novel multi-branch architecture that, for the first time, integrates the Joint Embedding Predictive Architecture (JEPA) into this domain. The method synergistically combines self-supervised semantic representations from a JEPA-pretrained Vision Transformer with supervised convolutional features from EfficientNetV2-S, enabling parameter-efficient complementary learning and multimodal feature fusion. Evaluated on the PVF-10 and ISM datasets, the model achieves F1 scores of 97.53% and 94.69%, respectively, while reducing the number of parameters by 47.2% compared to GEPFNet, thereby significantly enhancing both performance and computational efficiency.

Abstract

The rapid expansion of solar photovoltaic (PV) systems has increased the need for reliable and scalable fault classification, as manual inspection is impractical at scale. Thermal infrared (IR) imaging provides a non-contact solution for identifying PV faults; however, accurate classification remains challenging due to class imbalance, limited texture information, and subtle thermal differences. In this work, we investigate the applicability of Joint-Embedding Predictive Architecture (JEPA) for thermal IR PV fault classification across various scenarios and propose JEFFNet (JEPA-EFFicientNet), a multibranch architecture that combines JEPA-based self-supervised representation learning with EfficientNetV2-S-based supervised convolutional feature extraction. JEFFNet fuses semantic representations from a JEPA-pretrained Vision Transformer with convolutional features from EfficientNetV2-S, enabling complementary feature learning. JEFFNet is evaluated on two public thermal IR datasets, PVF-10 and InfraredSolarModules (ISM), for both multiclass and derived binary (healthy/faulty) classification. On PVF-10, JEFFNet achieves an F1-score of $93.21$ and an accuracy of $94.33$ in the 10-class task, and an F1-score of $97.53$ and an accuracy of $96.41$ in the derived 2-class task. On ISM, JEFFNet achieves an F1-score of $72.60$ and an accuracy of $83.88$ in the 12-class task, and an F1-score of $94.69$ and an accuracy of $94.78$ in the derived 2-class task. JEFFNet also uses only 108.6M parameters versus 205.91M for GEPFNet, a 47.2\% reduction. These results demonstrate that combining self-supervised semantic and supervised convolutional features provides an effective, parameter-efficient solution for thermal IR PV fault classification. The source code is publicly available at https://github.com/Azimi2kht/JEFFNet

Problem

• solar PV fault classification
• thermal infrared imaging
• class imbalance
• fault detection
• reliable scalable classification

Innovation

• JEPA
• self-supervised learning
• thermal infrared imaging
• fault classification
• multimodal feature fusion