Isolation and Detection of Arc Fault Noise in a Real PV System Using Current Demodulation and Autocorrelation Coefficients

The safety, reliability, and efficiency of photovoltaic systems hinge on effective arc fault identification and detection. This research presents a new online current demodulation algorithm to filter the switching signal injected into the DC side of the inverter. Therefore, the detection algorithm adapts performance in real-time so that it can accurately detect an arc fault without being confused with unknown inverter harmonics or other unwanted tripping events. This increases robustness of the arc fault detector. The paper provides deep theoretical analysis and explanation of the arc fault detection algorithm. It is also verified in 1) an experimental PV array using various commercial inverters with different PWM frequencies and 2) data captured by Sandia National Laboratories.