Abstract

The growing integration of photovoltaic systems into modern energy networks demands intelligent monitoring solutions to optimize performance and ensure long-term reliability. This study proposes a predictive framework based on an Adaptive Neuro - Fuzzy Inference System for estimating the power output of photovoltaic cells using electroluminescence images and current - voltage curves. The model extracts three key features from electroluminescence images counts of black, grey, and white pixels which reflect internal defects and degradation patterns. These image derived parameters are correlated with electrical performance indicators obtained from current voltage curves to enhance prediction accuracy. By combining the learning capabilities of neural networks with the interpretability of fuzzy logic, the adaptive neuro - fuzzy inference model provides accurate estimations of the normalized maximum power point while maintaining transparency through interpretable fuzzy rules. The system is implemented on a cloud - based platform, enabling real - time data analysis and offering valuable insights for performance assessment and quality control of photovoltaic cells. Experimental validation on a dataset of six hundred crystalline photovoltaic cells demonstrates high predictive accuracy, achieving a mean absolute error of 0.073 and a mean squared error of 0.0084. This approach enhances photovoltaic module efficiency evaluation and supports the development of more effective monitoring strategies in photovoltaic systems.