Application of Adaptive Neuro-Fuzzy Inference for Output Power Estimation in Photovoltaic Cells

Abstract This manuscript introduces two Adaptive Neuro-Fuzzy Inference Systems developed to predict the energy output of Photovoltaic cells. These models are trained using Electroluminescence imagery of the cells for input data along their Current-Voltage curves, which offer insights into the cells’ power output. The input characteristics of the cells are quantified based on the pixel distribution and categorized into three distinct classes: Black, White, and Gray values. By synergizing the rulebased interpretability of Fuzzy Logic with the adaptive learning prowess of Artificial Neural Networks, ANFIS is found as a superior approach for addressing this issue. Comparative analysis with other Machine Learning techniques demonstrates the ANFIS models’ enhanced performance, achieving a Mean Absolute Error (MAE) of 0.053 and a Mean Squared Error (MSE) of 0.007.