IEEE Power & Energy Magazine - May/June 2018 - 33

The application of incipient failure detection,
coupled with close to real-time communication,
will allow early corrective actions.

two feeders have equal ratings. extended statistical and data
mining analysis, along with the use of techniques such as pattern recognition, could be used to identify the root causes of
the differences between the two feeders as well as whether a
site visit or repair is necessary. essentially, this is an inputto-risk-based maintenance schedule, with a greater number
of anomalies located on the distribution system leading to a
greater potential for an outage.
data-driven techniques based on µPMu measurements
can sometimes turn traditionally challenging problems of
power distribution systems into somewhat trivial tasks. one
such example is phase identification, which is the problem
of identifying the correct phase connection for each single-,
two-, or three-phase load. traditionally, phase identification is performed by analyzing the correlations across voltage rms values. however, phase identification can also be
accomplished using a simple data-driven analysis of phase
angle measurements.
an example is given in figure 7. here, we plot the probability distribution, in the form of a discrete histogram, of the
relative phase angle difference between the voltage phasor at
a single-phase load and the voltage phasor at each phase of the
substation. the distribution is formed close to 0°, 240°, and
120° for phases a, B, and c, respectively. clearly, the load
is connected to phase a. this is also a technique that could
enhance the ability to use smart-meter data. With the meter's
phase identified, more focused control can be developed at a
granular level. this technique can be applied throughout the
distribution system and will become significantly important

140

120

80
60
40
20
0
-1

0
1
Relative Phase Angle (°)
(a)

100
80
60
40
20
0
236

Frequency

100

1,000

140
1,000

140

Frequency

1,000

statistical analysis tools to conduct such classification, in
large part, automatically.
as shown in figure 6, all transient voltage events above
0.1% rated magnitude during one day are cross-correlated
using synchronized voltage measurements on three phases of
two neighboring feeders. each point indicates an event that
was detected on at least one feeder. the location on the y-axis
indicates the transient voltage change detected on feeder 1,
while the location on the x-axis indicates the change detected on feeder 2. the events are either voltage sags or voltage
swells. In figure 6, and on each phase, the diagonal points
indicate those voltage transient events initiated at the transmission level. they comprise roughly one-third of all major
voltage transient events that appear on these two distribution feeders. they are not of primary interest to distribution engineers.
the events distribution engineers care about are, rather, the
two groups of points that fall inside the two dashed ellipsoids
marked in figure 6. note that the ellipsoids are marked on
only phase a to avoid crowding the figure; however, similar
ellipsoids can be also identified for phases B and c. "group
1" indicates the major transient voltage sags caused by the
equipment and/or loads on feeder 1. "group 2" indicates the
major transient voltage sags caused by the equipment and/
or loads on feeder 2. the events in group 1 comprise 65%,
71%, and 70% of all major voltage transient events on phases
a, B, and c of feeder 1, respectively. the events in group 2
comprise 60%, 65%, and 64% of all major voltage transient
events on phases a, B, and c of feeder 2, respectively. the

237
238
239
240
Relative Phase Angle (°)
(b)

100
80
60
40
20
0
116

117
118
119
120
Relative Phase Angle (°)
(c)

figure 7. The relative phase angle difference distribution between a single-phase load and each of the three phases at the substation: (a) phase A, (b) phase B, and (c) phase C.
may/june 2018

ieee power & energy magazine

Table of Contents for the Digital Edition of IEEE Power & Energy Magazine - May/June 2018