Enhancing photovoltaic cell classification through Mamdani Fuzzy logic: a comparative study with machine learning approaches employing electroluminescence images

Abstract

This study introduces a Mamdani Fuzzy Logic model designed to classify solar cells based on their energetic performance. The model incorporates three distinct inputs, namely the proportions of black pixels, gray pixels, and white pixels, extracted from Electroluminescence images of the cells. Additionally, an output is included to signal potential issues with input data. The development of the model involved utilizing cells with known performance, determined through the measurement of Intensity-Voltage Curves. The efficacy of the model was demonstrated through testing with a validation set, yielding an accuracy rate of 99.0% in the Polycrystalline dataset and 98% in the Monocrystalline. In comparison, traditional machine learning methods such as Ensemble Classifiers and Decision Trees achieved inferior accuracy rates. These results show the superior problem-solving capability of the presented Fuzzy Logic model over conventional machine-learning approaches.