Enhanced Online Fault Detection and Classification Method for PV Systems

This paper proposes a method for detecting and classifying faults in photovoltaic (PV) systems. The approach aims to simplify the fault detection process by monitoring the rate of change of power across the PV array, utilizing two distinct thresholds to identify and categorize faults based on severity. The proposed method eliminates the need for complex AI models, significantly reducing hardware requirements and offering scalability from small PV plants to large-scale power installations. Classification is achieved by analyzing the trajectory of power at the array level along with each string’s current, using a two-dimensional plane (X-Y plane) to distinguish different fault types. This technique provides clear identification of various fault scenarios, such as cell-to-cell, string-to-string, and pole-to-pole faults. The simulation results, obtained using MATLAB Simulink, demonstrate the accuracy and efficiency of the proposed approach, effectively differentiating between high and low impedance faults. Notably, the observed behavior of the power and current trajectories varies for different fault types, which aids in precise classification and localization. The technique relies on the degree of power loss, enabling clear distinctions between high and low severity faults. For instance, all solid connections exhibited a power change rate exceeding -100 p.u./s, while pole-to-pole faults, regardless of whether they were solid or had high fault resistance, were categorized as high severity. Conversely, partial shading within the strings was identified as a low severity fault. The method’s ability to reliably detect both high and low impedance faults sets it apart, making it a cost-effective and robust solution for enhancing PV system reliability.