Detecção de sujidade e sombreamento em painéis solares com inteligência artificial utilizando aplicativos móveis

Abstract

Solar energy has emerged as one of the main sources of renewable energy, given its abundance and potential for carbon emission reduction. However, the efficiency of solar panels can be compromised by various factors, among them dirt and shading. Early and accurate detection of these issues is crucial to maintain the operational efficiency of the panels and ensure a return on investment. With this goal in mind, this study developed a system for detecting dirt and shading on solar panels using advanced Artificial Intelligence techniques. Machine learning algorithms, such as the Light Gradient Boosting Machine (LGBM) and the Extra Trees Classifier (ETC), were employed to analyze the I-V curves and identify characteristics associated with dirt on the panels. The PyCaret tool was used to facilitate data processing and analysis. To complement the approach, two mobile apps were developed: one aimed at solar panel owners and another for maintenance companies. The machine learning algorithms demonstrated high accuracy in detecting dirt and shading, with LGBM standing out for its superior performance compared to ETC. The developed mobile apps proved to be practical and efficient, allowing users to monitor the status of their panels in real-time and request maintenance services when necessary. In conclusion, the combination of advanced data analysis techniques with mobile solutions offers an innovative approach to optimize the efficiency of solar panels, ensuring a more efficient and cost-effective operation.