Photovoltaic Array Fault Detection Using Ensemble Learning-Based Technique

Fault detection in the photovoltaic array aims to ensure a stable and continuous power supply. Detecting faults in photovoltaic arrays is challenging because normal and faulty conditions can sometimes exhibit similar characteristics. This paper presents an approach for photovoltaic array fault detection using an ensemble learning-based technique. A 3.2 kW MATLAB-Simulink photovoltaic array model is developed, and the fault characteristics of short circuit faults, line-ground faults, and hot spot faults are analyzed to identify the most suitable measurements for effective fault detection and classification. It is observed that photovoltaic array measurements such as voltage, current, power, rate of change of voltage and current over time (dV/dt and dI/dt), and change in power to voltage and current (dP/dV and dP/dI) exhibit distinct fault characteristics, making them valuable for accurate fault discrimination. The proposed model is trained using the selected photovoltaic array measurements, and its effectiveness is validated through a testing dataset, with performance indices derived from the confusion matrix. The proposed technique achieves a promising fault detection accuracy of 99.72%. Additionally, the performance of the proposed technique is evaluated and compared with other artificial intelligence-based techniques. The results demonstrate that the proposed method outperforms these alternatives.