Speaker
Description
The massive integration of new power
generation plants using decentralised renewable
energy resources (DER) implies an expansion of the
electrical distribution power grid with the construction
of more transmission lines and substations. With this
expansion, the probability of the occurrence of
permanent and intermittent faults will increase.
The time interval and duration of intermittent faults
vary in time, being characterised by a non-stationary
spectrum. Existing algorithms to detect permanent
faults may not be effective for the detection of
intermittent faults due to the time-varying spectrum
characteristics of the latter. The Intelligent Electronic
Devices (IEDs) in substations should include new
algorithms able to effectively detect intermittent faults.
This paper studies and tests the effectiveness of a low
processing time-frequency analysis methodology for
detecting intermittent faults in power distribution
grids with low impedance neutral. An analysis in time
of the cumulated zero-sequence energy to check for the
spike persistence is performed. Besides this, a
frequency analysis to check the persistence of spectrum
peaks around the natural oscillation frequency is also
performed. This frequency analysis is important to
distinguish from harmonics generated by power
electronic converters. Low processing frequency
analysis techniques as Hilbert Transform (HT) and
Derivative Sign Variation (DSV) are adopted. The
distribution power grid topology and adopted neutral
regime influence the natural frequency of oscillation. A
reference distribution grid topology with a high-to
medium voltage distribution transformer and two
secondary medium-voltage feeders, one healthy and
the other with an intermittent fault, is considered. The
grid with a low impedance neutral using a zigzag (ZZ)
transformer for the neutral grounding.