Windowed standard deviation-based transient feature extraction for earth fault protection in MV networks

In this paper, the windowed standard deviation (WSTD) is used to extract the initial transients created in residual current waveforms generated due to earth faults in medium voltage (MV) compensated networks. The sliding window width used for computing the standard deviation is evaluated from the frequency filter point of view. The fault cases occurring in a compensated 20 kV network are simulated by ATP/EMTP concerning practical fault case such as arcing faults. Results provide evidence that the windowed standard deviation can be used to extract initial transients although they are non-stationary sinusoidal signals.