Signal Processing - March 2017 - 17

Wireless
Sensors

Access
Point

Data

Cloud Computing
Server Farm

Internet
Connection
Operation Center

Cloud
Network

Cloud Data
Center

Distributed Electricity
Generation

Transmission
Network

Smart City
Smart Home

Smart Metering and
Consumers' Demands
Information Collection

Cloud-Based Real-Time
Smart Grid Auditing
and Anomaly Detection

Figure 3. Binghamton University researchers have proposed a new SSA algorithm-based model for remotely detecting and locating power grid faults.
This diagram shows how a cloud-based monitoring system using the model might be configured. (Image courtesy of Binghamton University.)

from the Department of Electrical and
Computer Engineering, and doctoral student Zekun Yan from the Department of
Electrical and Computer Engineering.
The SSA-based change point detection
method is a nonmodel and structure-free
data analysis technique, Prof. Chen says.
"As a result, on one hand, theoretically, it
can detect faults of any category because
SSA don't have a limitation for data structure." On the other hand, Prof. Chen notes,
history records and prior information
about the fault condition are not needed to
build the model. "In addition, experimental results have shown high accuracy for
this method," Prof. Chen says.
At the individual local sensor level,
the moment a disturbance arrives at
the measured position is a critical factor in disturbance detection. "With the
help of synchronization, the disturbance
record data are marked by time instance
such that it can be compared with other
records," Prof. Chen says. "Therefore,
naturally, it's a time series and suitable for
a corresponding data analysis technique."
The new approach utilizes a traveling
wave-based approach that's both simple
and well suited for use in the wide-area
systems, such as power grids. "The trav-

eling wave-based method focuses on the
waves generated by the disturbance and
the correlation between the forward and
backward wave arrival or direct detection of the arrival time on a single line,"
Prof. Chen says. Specifically, the approach
uses a time difference of arrival (TDOA)
method. "The TDOA method assumes
a unified wave traveling velocity and utilizes the different arrival times at different
sensors to locate the wave start point-the
fault source," Prof. Chen says. "The critical
challenges for the method are the arrival
time detection accuracy and the wave traveling velocity error."
Developing a highly accurate power
grid detection/location technology poses
some unique challenges. "First, the propagation speeds of the electromechanical
signals vary from 100 miles to 1,000
miles per second among different paths
in the power grid due to the adoption of
variant transmission technologies," Prof.
Chen says. "Second, absolute geological
distance is used to simplify the calculation in spite of the fact that the electromechanical waves propagate along with the
power grid topology."
Although the individual measurement
data has propagation error and modifica-

IEEE Signal Processing Magazine

March 2017

tion error, a proposed temporal scanning
algorithm would take advantage of the
abundant amount of information supplied
by frequency disturbance recorders dispersed across the grid to reach an accurate value. "According to the law of large
number, with the large number of samples
the estimate will gradually converge to the
true value," Prof. Chen says.
The researchers continue to refine
the system and are currently addressing
two key issues: the fast detection of weak
oscillations and the prediction of potential vibrations based on the detection of
disturbances and oscillations.
Meanwhile, the current SSA algorithm can only detect and locate problems; it cannot predict what may happen
in the future. The researchers hope that
the current platform's ability to detect
subtle power grid change changes could
someday be improved to the point where
it will be able to forecast problems before
they occur.

Author
John Edwards ( je dwa rd s@joh n
edwardsmedia.com) is a technology writer based in the Phoenix, Arizona, area.
SP

http://www.edwardsmedia.com

Table of Contents for the Digital Edition of Signal Processing - March 2017