Fault detection in photovoltaic systems using unmanned aerial vehicle-captured images and rough set theory

Abstract

The growing reliance on photovoltaic (PV) systems as a sustainable energy source is challenged by performance degradation due to faults, necessitating efficient fault detection methods. This study proposes an AI-driven approach using unmanned aerial vehicle (UAV)-captured images for automated PV module inspection. Advanced feature extraction techniques, including Texture Analysis, Fast Fourier Transform (FFT), Grey-Level Co-occurrence Matrix (GLCM), Grey-Level Difference Method (GLDM), and Discrete Wavelet Transform (DWT), were employed to analyze image data. A Rough Set-Based Rule Classifier was optimized, achieving 100% accuracy when paired with DWT features. Additionally, data augmentation techniques were integrated to enhance model robustness. The proposed method improves PV system maintenance by enabling precise, non-destructive fault detection, ensuring higher efficiency and reliability for solar energy adoption.