Feature Enhancement Method for Weak Photovoltaic Series Arc Fault Signals

For intensely burning photovoltaic (PV) series arc faults without strong noise interferences, their time-frequency features are easily to be discovered. However, various PV systems would generate noise interferences to the arc fault signal, causing difficulties to distinguish the arc fault and normal states. To solve this kind of problem, new measurements should be taken to acquire obvious arc fault features even from the weak arc fault electrical signals.In this paper, weak PV series arc fault electrical signals are acquired from the designed experimental setup with different load types firstly. Then it is found that the performance of arc fault features are not that satisfying in higher frequency bands after directly applying the existing Db9 based wavelet transform, causing the arc fault detection problem. Next, arc fault features are enhanced in most frequency bands by conducting the proposed Rbio3.1 based wavelet transform. Finally, the stochastic resonance (SR) method is proposed to further enhance Rbio3.1-based arc fault feature. The compared results prove that the combination between SR method and Rbio3.1 wavelet transform show the effective feature enhancement ability facing weak PV series arc fault electrical signals with different inverters and resistors.