Fault Identification and Classification of Solar PV System

There are numerous intrinsic faults in solar photovoltaic (PV) systems that reduce their efficiency and energy output. Accurately classifying and promptly identifying faults are important components of ensuring a system’s dependability and effectiveness. In this study, solar PV module faults are classified and identified using CNN and SVM. CNN detects faults with little preprocessing by using deep learning to extract spatial information from solar images. SVM, on the other hand, uses feature sets that have been retrieved to categorize defects through statistical learning. The classification accuracy, computational complexity, and stability despite environmental changes of both methods are compared. According to experimental results, SVM provides consistent results with reduced computing cost for structured data, whereas CNN delivers higher accuracy because it can immediately learn complex patterns from images. According to comparative analysis, $C N N$ is a better choice for real-time monitoring of large-scale solar PV systems since it performs SVM in identifying small problems like electrical, physical, bird drop, and snowfall. This study highlights how AI-based diagnosis might increase solar power systems’ lifespan and efficiency.