Series PV Arc Fault Detection using Current Demodulation and Autocorrelation Coefficients

Traditional arc fault detection methods for series photovoltaic (PV) systems often require specific tuning of the arcing frequency detection bandwidth to avoid false alarms from switching noise of the PV inverter or even noise from the loads. This may require a priori knowledge of the PV inverter or advanced sensing features. However, this research combines using adaptation and frequency detection with current demodulation. Then advanced pink noise models of the series PV arc faults are characterized, eliminating false alarms from the inverter and other noise sources, while removing the need to know performance specification details of the inverter.