Fault Detection in Photovoltaic System Through Machine Learning Methods

Implementing an accurate fault detection and estimation system for photovoltaic arrays is crucial for improving the safety and dependability of solar power systems, aiding in renewable energy transition and sustainable future. The study uses artificial intelligence algorithms to quantify and identify electrical issues in photovoltaic (PV) farms. Among the techniques used are fine gaussian support vector machine (SVM), medium gaussian SVM, quadratric SVM, ensemble bagged trees, ensemble subspace discriminant, and linear SVM. The approaches are evaluated using a variety of metrics, such as the area under the receiver-operating curve, F-measure, specification, recall, precision, and classification accuracy. These techniques are applied on a 250 kW grid-connected photovoltaic farm with multiple faults scenario. The results demonstrate that ensemble bagged trees achieved the highest performance with an accuracy of 97.2% and F1-score of 96.8%, while medium gaussian SVM showed the lowest performance with 80% accuracy. Overall, the proposed approach confirms the feasibility of applying machine learning to robust PV fault detection and provides a benchmark comparison across multiple classifiers.