Machine Learning and Thresholding Approach for Defects Classification in Solar Panels

Abstract

This research addresses critical aspects of solar photovoltaic (PV) system maintenance and monitoring to ensure sustained performance. Emphasizing solar panel reliability, the study employs image processing, clustering algorithms, and machine learning (K-Means, Naive Bayes) to detect and categorize factors impacting efficiency, such as dust accumulation and sunlight exposure. The developed system facilitates comprehensive assessment and classification, enhancing operational lifespan. Demonstrating versatility, the project incorporates alternative feature extraction and interactive threshold selection, ensuring adaptability to diverse scenarios. Experimental validation, including hotspot detection in thermal images, underscores the robustness of the proposed methodology, contributing significantly to solar panel monitoring and maintenance advancements. © The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2025.