A Synthetic Method for Photovoltaic Series Arc Fault Detection with Different Loads and Its Identification

Series arc fault that frequently occurs in photovoltaic (PV) generation system greatly reduces safety, economy and stability level of the operating system. However, factors as time-varying PV source caused by weather, outside electromagnetic interference especially from the inverter make PV series arc fault more complex to be detected. This paper aims at providing a relatively reliable method for a new smart device called arc fault circuit interrupter (AFCI) to prevent PV system from threats by arc fault. In this paper, normal electric signals, fault electric signals and fault-like electric signals have been acquired via an experimental platform with PV system and different loads. Bringing PV series arc fault into the system, fault electric signals appear transient-state stage and steady-state stage which is also adjusted by inverter working mode. Fault-like electric signals have nearly the same transient process as fault electric signals. Euler method based on electric change and short time Fourier transformation (STFT) based on harmonic change are chosen to diagnose this kind of fault. A relatively satisfying synthetic algorithm based on a detection variable from each method has been proposed to identify fault conditions regardless of interference by fault-like conditions.