Fault Diagnosis in Photovoltaic Systems Using Machine Learning Algorithms

Driven by environmental concerns, the global energy landscape is undergoing a significant shift from use of fossil fuels to renewable energy sources. Solar energy, particularly through photovoltaic (PV) technology, has emerged as a prominent renewable energy source. However, use of PV systems faces challenges due to fault occurrences, which negatively impact their efficiency and power output. With machine learning revolutionizing the energy sector, expectations hold in offering potential solutions to fault diagnosis challenges in PV systems. This study focuses on utilizing machine learning algorithms (MLA) for fault diagnosis in PV systems. Through a MATLAB simulation of a 7.5 kW PV system, three fault scenarios were implemented to generate the dataset. MLAs were trained and tested using MATLAB classification application, with cross and hold-out validation techniques employed for model validation. Results revealed that the wide neural network algorithm achieved the highest accuracy, reaching 96.30% during training and 98.39% during testing with the cross validation of 10 folds. The ensemble algorithms also demonstrated promising results, achieving accuracies of 93.73% and 94.06% during training and testing, respectively, with cross-validation. These findings underscore the effectiveness of machine learning algorithms in accurately diagnosing faults in PV systems, offering valuable insights for PV system maintenance to ensure efficiency, reliability, availability, and overall performance.