Detection of Hot Spots on Photovoltaic Modules in a Solar Power Plant by Measuring the Dissimilarity of Thermal Images Taken by Drone
Abstract
Early identification of thermal anomalies, such as hotspots, is essential to ensure the efficiency and longevity of photovoltaic systems. This work presents a methodology based on aerial inspection using a drone and quantitative analysis of thermal images for hotspot detection in photovoltaic modules. The approach combines module segmentation, projective correction, extraction of temperature matrices through the DJI Thermal SDK , and the application of dissimilarity metrics — sum of absolute differences and sum of squared differences. The inspections were carried out at a 1 MWp solar power plant located in the State of Sergipe, Brazil, with thermal data acquisition performed in accordance with IEC TS 62446-3:2017. The results show that the applied metrics effectively distinguish modules with and without hotspots, even under controlled induced-shading conditions. The proposed methodology demonstrated technical feasibility for fast, safe, and accurate inspections, contributing to predictive maintenance in large-scale photovoltaic systems. This study reinforces the role of aerial thermography as a strategic tool for the efficient and sustainable operation of solar power plants.
Keywords
Photovoltaic modules, drone thermography, hotspots, thermal dissimilarity, predictive maintenance