1

Line Outage Detection and Localization via

Synchrophasor Measurement
Xianda Deng2, Desong Bian1, Di Shi 1, Wenxuan Yao2,3, Zhihao Jiang2, Yilu Liu 2,3
1
Global Energy Interconnection Research Institute North America
2
University of Tennessee Knoxville, TN, 3Oak Ridge National Lab
xdeng6@vols.utk.edu, desong.bian@geirina.net, di.shi@geirina.net,
wyao3@ vols.utk.edu, zjiang13@vols.utk.edu, liu@utk.edu

Abstract—Since transmission lines are crucial links in the such as synchrophasor measurements, advanced metering
power system, one line outage event may bring about interruption infrastructure, and home energy management system.
or even cascading failure of the power system. If a quick and Nowadays, the density of synchrophasor is increasing
accurate line outage detection and localization can be achieved, the
dramatically to observe the dynamic behavior of the system
system operator can take necessary actions in time to mitigate the
negative impact. Therefore, the objective of this paper is to study following a contingency, which gives the unpreceded insights
a method for line outage detection and localization via to the system [23]-[28]. For example, there are 114 and 238
synchrophasor measurements. The density of deployed Phasor PMUs installed at Jiangsu power grid, respectively. As shown
Measurement Units (PMUs) is increasing recently, which greatly in Fig. 1, the density of distributed PMU is significant high,
improves the visibility of the power grid. Taking advantage of the which covers all 500kV transmission lines and parts of 220 kV
high-resolution synchrophasor data, the proposed method utilizes
transmission lines with the reporting rate 25 Hz. To utilize the
frequency measurement for line outage detection and power
change for localization. The procedure of the proposed method is PMU data and achieve wide area monitoring purpose, a PMU
given. Compared with conventional methods, it does not require based situational awareness data analytics platform has been
the pre-knowledge on the system. Simulation study validates the developed by Global Energy Interconnection Research Institute
effectiveness of the proposed method. North America (GEIRINA) [29]-[33]. The PMU based
Index Terms— Detection and localization, Line outage, PMU, platform collects synchrophasor measurements with massive
synchrophasor channels in real time from Jiangsu power grids and processes
I. INTRODUCTION large amount of data, which can be affected by latency from
PMU device or communication network. The platform not only
T he detection and localization of transmission line outage in
power system is of great significance for the system
operators to take prompt action to avoid the widespread damage
incorporates event detection application developed by
GEIRINA, but also provide interface for event detection
applications from third party. The density synchrophasor
and maintain the reliability of power supply[1]-[5]. Most of
measurement brings the opportunity to detect the line outage
current methods rely on angle data along with network
location and locate the fault location without knowing the
susceptance matrix to calculate power injection change, which
system parameters. The line outage detection approach
is a high computation burden and requires the information of
introduced by [34] was employed in the PMU based platform
system parameter [2], [6]-[11]. The method using PMU angle
and the reported line outage locations have significant deviation
data and network susceptance was originally proposed in [2]
from actual line outage location. Similar phenomenon was also
and pre- and post-outage power flow were calculated to match
founded in simulation cases in New England ISO (ISO-NE) and
the measured event. Later, compressive sensing and global
Tennessee Valley Authority (TVA) systems.
optimization techniques are proposed to improve the method in
To address the issue mentioned above, this paper focuses on
[7] and [8]. A general Bayesian criterion was employed to
the method for line outage detection and localization via
handle the uncertainty issue of PMU data in [9]. Different new
synchrophasor measurement. First, the line outage is detected
schema and frame are developed to deal with bad PMU
via employing low pass filter and peak detector on
measurements in [10] and [11]. However, the system parameter
synchronized frequency measurements. Once a line outage
may not be available all the time due to strict security concerns.
event is triggered, the location of the fault line will be
With the rapid transformation from traditional power system
pinpointed using power flow change. The requirement of
into the smart grid, there are various types of novel applications
computational effort for the whole process is not high thus
involved into the system [12]-[15]. These smart grid
outage location can be estimated in real time. The proposed
applications are relying on the high quality synchronized data
method is straightforward and easy to implement. It also can be
and advanced communication network infrastructure [16]-[24],
used for cross-checking line outage event via SCADA.
This work is supported by the CURENT Industry Partnership Program and the The rest of the paper is organized as follows. In Section II,
Engineering Research Center Program of the National Science Foundation and the proposed line outage detection and localization method via
DOE under NSF Award Number EEC-1041877.
 2

synchrophasor measurement are given. In Section III, the ∆𝑃𝑘 φ

ς𝑚 = 𝑃𝑚
= 1−φ𝑘 (2)
characteristic of power flow change redistribution during line 𝑚
outage is explored via PSS/E simulation on models of ISO-NE where ∆𝑃𝑘 is the power flow transfer changes at line k and 𝑃𝑚
and TVA systems. The simulation study on the ISO-NE model is the of pre-outage real power flow at line m. According to the
is conducted in Section IV. The conclusions and future work definition in Eq.(1) and (2), PDTF and LODFs are less than 1.
are drawn in Section V. Meanwhile, bus in pre-and post-outage condition must
follow Gustav Kirchhoff’s Current Law (KCL). Defining the
power flow change at terminal k1 is ∆𝑃𝑘1 , the power flow
change outage follows:
∆𝑃𝑘1 = Sum(∆𝑃𝑗 ), (3)
𝑠. 𝑡. , 𝑗 ∈ 𝑏𝑟𝑎𝑛𝑐ℎ 𝑐𝑜𝑛𝑛𝑒𝑐𝑡𝑒𝑑 𝑡𝑜 𝑘1
where 𝑗 is the indexes of the lines.
In an actual power system, the disturbance usually spread out
from the source to the rest of the system. As a result, the bus
with relatively large power flow changes might be closer to the
location of the outage line, that is ∆𝑃𝑘1 > ∆𝑃𝑗1 when distance
of 𝑘1 to the outage location is smaller than 𝑗1 . Therefore, the
bus of the outage line is likely to have the largest power change
in the system.
Using the active power change from synchrophasor, the
location of outage can be estimated. Specifically, once a line
outage event is detected, noise in active power measurement is
Fig. 1. Map of synchrophasor deployment in Jiangsu power grid [29] filtered by a median filter. With detected event timestamp, the
active power change between pre-outage and the post-outage is
II. LINE OUTAGE DETECTION AND LOCALIZATION calculated with the filtered active power measurements. By
The proposed method for line outage detection and location ranking the active power change on the monitored transmission
estimation in the paper includes two steps: (1). Line outage lines, the location of the outage line can be determined with
event detection using frequency measurement; (2). Outage line maximum value. The process of the line outage detection and
localization using active power measurement. localization method is presented in Fig. 2.
The frequency measurements from deployed PMUs is used -
Raw LP filter LP filter Peak
to monitor that if there is ongoing a line outage event in the frequency (median) (mean) detection
system. The principle of outage detection can be found in Ref. -
[34]-[36]. The frequency measurements are first fed into a +
moving median filter to remove random spikes and high-
frequency noises. After that, a moving mean filer is used to Line outage
extract the frequency trend as a reference. Then de-trended report
frequency data is subtracted frequency trend from filtered
frequency measurements. Two thresholds are empirically set
based on statistical analysis on historical data [34]. The event
Power flow Estimate trip
outage will be triggered using threshold evaluation. The event Raw power LP filter
change outage
time can be determined via GPS timestamp on the flow (median)
caculation location
measurements [37]-[40], which will be utilized further for event
location estimation with active power measurements. Fig. 2. Flowchart of line outage detection and localization
When a line outage event happens, the active power flow
will be redistributed partially since the power flow on the III. DISTRIBUTION OF POWER CHANGE IN TVA AND ISO-NE
tripped line has to transfer the rest of the system abruptly, which This section investigates the characteristic of power flow
provides useful information for line outage location estimation. change distribution caused by line outage via PSS/E simulation.
The Power Transfer Distribution Factors (PTDFs) of line l Line outage events are simulated in both ISO-NE and TVA
respect to a power flow transaction ∆𝑤 in a lossless model is systems, respectively. ISO-NE system consists of 3447 buses
defined as [41]-[44]: and 2479 branches in 71,992 mile². The total generation is
∆𝑙 18.1GW and total load is 21.8 GW in the system. There are 16
φ𝑙 = ∆𝑤 (1)
tie lines, which carry 3.7 GW power flow, connecting to the
where ∆𝑤 is MW of power transfer between two location and system. TVA system has 1920 buses and 1720 branches. There
∆𝑙 is the power transfer via branch l respect to the transaction. are 28.1 GW generation and 31.6 GW load within TVA system.
Then, for an outage at line m, Line Outage Distribution Factor The TVA system are connected with external system via 70 tie
(LODF) is defined as the portion of pre-outage real power flow lines and total 3.5 GW energy are delivered by the tie lies. The
transfer to a line k [41], which can be represented as
 3

simplified system diagram of ISO-NE and TVA systems are Fig. 6. It can be seen that the power flow changes at the
showed in Fig. 3 and Fig. 4, respectively. terminals of the outage line have the highest value for the both
cases. What is more, the power flow change closer the outage
line generally has a larger value than the line far away from the
event location.

Fig. 3. ISO-NE model—transmission network map [45]

(a)

Fig. 4. Diagram for Tennessee Valley Authority [46]

TABLE I. ERROR ANALYSIS FOR LINE OUTAGE LOCATION ESTIMATION

Power (b)
Termial1 Termial2 Voltage
System Case name flow Fig. 5. Distribution of power change in ISO-NE system (a) Line1 outage (b)
(Lat,Lon) (Lat,Lon) level(kV)
(MW) Line2 outage

Line1 41.51,-72.56 41.29,-72.90 407.36 345

ISO-NE
Line2 42.63,-71.05 42.70,-70.87 100.75 115

Line3 35.10,-85.02 34.05,-85.08 856.78 500

TVA
Line4 37.78,-86.48 37.26,-86.98 246.72 161

For the purpose of a comprehensive study, 4 transmission

lines in ISO-NE and TVA system are selected and tripped. The
voltages levels of the outage line are from 115 kV to 500 kV.
The terminal locations (latitude and longitude) of outage line
and pre-outage real power flow on the lines are given in Table I.
The locations of the disturbance and distribution of power flow (a)
changes caused by the line outage are shown from Fig. 5 to
 4

TABLE III. RESULT OF LINE OUTAGE LOCATION ESTIMATION

Max Average
Voltage Monitored Cases
error error
level by PMU numbers
(Mile) (Mile)

345 kV Y 37 0 0

230 kV Y 20 0 0

230 kV N 8 13.72 6.32

115 kV N 30 82.49 10.42

TABLE IV. PERFORMANCE COMPARISON FOR LOCATION ESTIMATION

Power Power Max
Case Voltage Monitored
flow change freq.
name level by PMU
(b) (MW) (mile) (mile)

Fig. 6. Distribution of power change in TVA system (a) Line3 outage (b) 1 345 kV Y 725.17 0 93
Line4 outage
2 230 kV Y 224.43 0 123
IV. SIMULATION STUDY
To verify the effectiveness of the proposed method for line 3 230 kV N 285.65 9.366 126
outage detection and localization, a simulation is conducted in
4 115 kV N 100.45 0 18
ISO-NE which assumes that synchrophasor covers all 345 kV
and part of (26%) 230 kV transmission lines. Line outages
events are triggered in PSS/E to evaluate the performance of the
events at different voltage levels. The parameters of the filter
for line outage detection are selected based on [34], which are
listed in Table II. The event detection module is implemented
in C#, while location module is developed by MATLAB. The
simulation tests are on a computer running a 64-bit Windows
10, with a 3.60 GHz Intel I7-7700U CPU and 16 GB memory.
TABLE II. PARAMETER SELECTION FOR LINE OUTAGE DETECTION

Parameters Values

Median filter window 7 points

Mean filter size 31 points

Detection window 20 points Fig. 7. Comparison of line outage localization (case 1)

First peak threshold 0.0045 Hz

Second peak threshold 0.0025 Hz

The line outage events can be successfully detected, and

event time can be accurately recorded for all simulation cases.
The location estimation of the events is further analyzed. The
information of the line outage location and estimation error are
in Table III. As shown in the Table III, the proposed method is
able to identify the outage line location precisely, when the
outage lines are monitored by synchrophasor. For the outage
lines without synchrophasor monitoring (115 kV), the
estimated location is close to actual outage line terminals. For
the cases with a large error in 115 kV case, the actual outage
lines are far away from PMU locations and the reported PMU
is the closest location to the actual outage line terminals.
Fig. 8. Comparison of line outage localization (case 2)
 5

reasonable accuracy. It can works as an effective tool for real-

time line outage detection and localization.
Simulation results manifest that the proposed approach is
promising for line outage detection and localization in large-
scale power system. The performances of the proposed
approach have not been validated with line outage events from
real power grid. Additionally, the approach has not been fully
tested for real-time implementation. Following are some future
works for further development:
(1) Validate the proposed approached with confirmed line
outage event from power grid utilities.
(2) Test the robustness of the approach with synchrophasor
measurements with low quality data.
(3) Develop interface with available PMU data platform and
Fig. 9. Comparison of line outage localization (case 3) evaluate the performance of this approach with real
synchrophasor measurements
VI. REFERENCES
[1] J. Chen, Y. Zhao, A. Goldsmith, and H. V. Poor, "Line outage detection
in power transmission networks via message passing algorithms," Signals,
Systems and Computers, 2014 48th Asilomar Conference, 2014, pp. 350-
354.
[2] J. E. Tate and T. J. Overbye, "Line Outage Detection Using Phasor Angle
Measurements," IEEE Trans. Power Systems, vol. 23, no. 4, pp. 1644-
1652, Nov. 2008.
[3] K. Dave, N. Mohan, X. Deng, R. Gorur and R. Olsen, "Analyzing
techniques for increasing power transfer in the electric grid," 2012 North
American Power Symposium (NAPS), Champaign, IL, 2012, pp. 1-6.
[4] X. Deng, "Exploring Six-Phase Transmission Lines for Increasing Power
Transfer With Limited Right of Way " M.Sc. Thesis, Arizona State
University, 2012
[5] X. Deng, Y. Chen and Ying Li, "Study on the CIM based data integration
platform," ISGT 2011, Hilton Anaheim, CA, 2011, pp. 1-5.
[6] J. Zhao, J. Tan, L. Wu, L. Zhan, W. Yao, Y. Liu, J. Gracia, P. Ewing,
“PMU error impact on measurement-based applications”, 2017 IEEE PES
General Meeting, Chicago, IL, 2017.
Fig. 10. Comparison of line outage localization (case 4) [7] ] H. Zhu and G. B. Giannakis, “Sparse overcomplete representations for
efficient identification of power line outages,” IEEE Trans. Power Syst.,
For the purpose of comparison, the locations of four line vol. 27, no. 4, pp. 2215–2224, Nov. 2012.
outage cases from 115 kV and 345 kV are estimated by the [8] J.-C. Chen, W.-T. Li, C.-K. Wen, J.-H. Teng, and P. Ting, “Efficient
identification method for power line outages in the smart power grid,”
proposed methods and the traditional methods using the IEEE Trans. Power Syst., vol. 29, no. 4, pp. 1788–1800, Jul. 2014.
maximum frequency magnitude change in Ref.[34] and [47]. [9] C. Chen, J. Wang and H. Zhu, "Effects of Phasor Measurement
The estimation errors for each case are given in Table IV. The Uncertainty on Power Line Outage Detection," IEEE Journal of Selected
estimation locations and actual line outage for each case are Topics in Signal Processing, vol. 8, no. 6, pp. 1127-1139, Dec. 2014.
[10] A., Mehebub, M, Biswaranjan, S., Siddhartha, "A New Approach of
plotted from Fig. 7 to Fig. 10. As shown in these figures, Multiple Line Outage Identification Using Phasor Measurement Unit
distances between the estimated locations by proposed methods (PMU) with Bad Data," 2018 International Conference on Current Trends
and the actual location of outage line are small while the towards Converging Technologies.
estimated location by methods using frequency changes has [11] W. Li, C. Wen, J. Chen, K. Wong, J. Teng and C. Yuen, "Location
Identification of Power Line Outages Using PMU Measurements With
significant deviations from actual outage locations. Bad Data," IEEE Transactions on Power Systems, vol. 31, no. 5, pp. 3624-
3635, Sept. 2016.
V. CONCLUSION AND FUTURE WORKS [12] D. Bian, M. Pipattanasomporn and S. Rahman, "A Human Expert-Based
Approach to Electrical Peak Demand Management," in IEEE
Awareness of line outage event and its location is critical to
Transactions on Power Delivery, vol. 30, no. 3, pp. 1119-1127, June 2015.
prevent cascading outages in today's modern power system. [13] Z. Liang, D. Bian, X. Zhang, D. Shi, R. Diao, Z. Wang, "Optimal energy
This paper presents a fast line outage detection and localization management for commercial buildings considering comprehensive
method utilizing synchrophasor measurements. The line outage comfort levels in a retail electricity market," Applied Energy, Volume
236, 2019, pp. 916-926.
is first detected via a peak detector on synchronized frequency [14] Y. Yan, D. Shi, D. Bian, B. Huang, Z. Yi and Z. Wang, "Small-signal
measurements, and location of the fault line is directly Stability Analysis and Performance Evaluation of Microgrids under
estimated via active power flow change. The proposed is Distributed Control," in IEEE Transactions on Smart Grid, 2018, pp. 1-1.
[15] Z. Liang, D. Bian, D. Su, R. Diao, D. Shi, Z. Wang, and W. Su, "Adaptive
straightforward and does not need the pre-knowledge on system Robust Energy Management Strategy for Campus-Based Commercial
topology and parameters. The feature of active power change Buildings Considering Comprehensive Comfort Levels," in IEEE PES
distribution caused by line outage is explored in both TVA and General Meeting (GM), Atlanta, GA, 2019.
[16] D. Shi, X. Chen, Z. Wang, X. Zhang, Z. Yu, X. Wang, D. Bian, "A
ISO-NE system. A comprehensive simulation study in ISO-NE Distributed Cooperative Control Framework for Synchronized
shows the method can precisely identify the outage line with
 6

Reconnection of a Multi-Bus Microgrid," in IEEE Transactions on Smart System Synchronized Measurement Devices, IEEE Transactions on
Grid, vol. 9, no. 6, pp. 6646-6655, Nov. 2018. Smart Grid, vol. 9, no. 2, pp. 1141-1149, March 2018.
[17] D. Bian, M. Kuzlu, M. Pipattanasomporn, S. Rahman and Y. Wu, "Real- [38]W. Yao, D. Zhou L. Zhan, Y. Liu, Y Cui, S. You and Y. Liu, GPS signal
time co-simulation platform using OPAL-RT and OPNET for analyzing loss in the wide area monitoring system: Prevalence, impact, and solution,
smart grid performance," 2015 IEEE Power & Energy Society General Electric Power Systems Research, vol.147, pp. 254-262, 2017
Meeting, Denver, CO, 2015, pp. 1-5. [39]X. Deng, , H. Li, W. Yu, W. Wang, Y. Liu, " Frequency Observations and
[18] D. Bian, M. Kuzlu, M. Pipattanasomporn, S. Rahman, D. Shi, Statistic Analysis of Worldwide Main Power Grids Using FNET/GridEye, "
"Performance Evaluation of Communication Technologies and Network 2019 IEEE Power & Energy Society General Meeting, Atlanta, GA.
Structure for Smart Grid Applications," IET Communications, accepted, [40]J. Zhao, S. You, H. Yin, J. Tan, Y. Liu, “Data quality analysis and
2019. solutions for distribution-level PMUs”, 2019 IEEE PES General Meeting,
[19] D. Bian, M. Kuzlu, M. Pipattanasomporn, S. Rahman, "Analysis of Atlanta, GA, 2019.
Communication Schemes for Advanced Metering Infrastructure (AMI)," [41]A. Wood and B. Wollenberg, Power Generation Operation and Control,
in IEEE PES General Meeting, National Harbor, MD, 2014. 2nd ed. New York: Wiley, 1996, p. 422.
[20] D. Bian, M. Kuzlu, M. Pipattanasomporn, S. Rahman, "Assessment of [42]H. Liu, J. Guo, W. Yu, T. Xia, R. Sun, M. Gardner, L. Zhu, Y. Liu. “The
Communication Technologies for a Home Energy Management System," Design and Implementation of the Enterprise Level Data Platform and Big
in IEEE PES Innovative Smart Grid Technologies Conference (ISGT), Data Driven Applications and Analytics”, 2016 IEEE PES Transmission
Washington, D.C., 2014. & Distribution Conference & Exposition, 2015.
[21] D. Bian, "A Novel Space-Time System Based on Turbo Channel [43]Z. Wang, D.Wang and et al, "Deep reinforcement learning of cell
Estimation Method," in International Conference on Wireless movement in the early stage of C. elegans embryogenesis,"
Communications, Networking and Mobile Computing, Dalian, China, Bioinformatics, 2018, 1, p.9.
2008. [44]T. Guler, G. Gross and M. Liu, "Generalized Line Outage Distribution
[22] D. Bian, D. Shi, M. Pipattanasomporn, M. Kuzlu and S. Rahman, Factors," in IEEE Transactions on Power Systems, vol. 22, no. 2, pp. 879-
"Mitigating the Impact of Renewable Variability with Demand-Side 881, May 2007.
Resources Considering Communication and Cyber Security Limitations," [45] ISO-NE model—transmission network map, available at:
in IEEE Access, vol. 7, pp. 1379-1389, 2019. https://www.nrel.gov/docs/fy14osti/61824.pdf
[23] D. Bian, H. Latchman, "High Speed Powerline Communications: State [46]Tennessee Valley Authority.electric infrastructure, avaialble at:
of the Art and Beyond," in International Multi-Conference on https://www.eia.gov/todayinenergy/detail.php?id=11151
Complexity, Informatics and Cybernetics (IMCIC), Tampa, FL, 2011. [47]Y. Song, W. Wang, Z. Zhang, H. Qi and Y. Liu, "Multiple Event
[24] Y. Meng, Z. Yu, D. Shi, D. Bian and Z. Wang, "Forced Oscillation Detection and Recognition for Large-Scale Power Systems Through
Source Location via Multivariate Time Series Classification," 2018 Cluster-Based Sparse Coding," IEEE Trans. Power Systems, vol. 32, no.
IEEE/PES Transmission and Distribution Conference and Exposition 6, pp. 4199-4210, Nov. 2017.
(T&D), Denver, CO, 2018, pp. 1-5.
[25] Y. Liu et al., "A Distribution Level Wide Area Monitoring System for
the Electric Power Grid–FNET/GridEye," in IEEE Access, vol. 5, pp.
2329-2338, 2017.
[26] C. Li, W. Zhou, and Sh. Yuan. "Iris recognition based on a novel
variation of local binary pattern." The Visual Computer 31.10 (2015):
1419-1429.
[27]Z. Jiang, L. Zhang, Z. Ren, Y. Chen, S. Huang, “Coupled analysis of
energy flow in a multiple energy system”, 2017 IEEE Conference on
Energy Internet and Energy System Integration (EI2), 26-28 Nov. 2017
[28]H. Liu, L. Zhu, Z. Pan, and et al. “ARMAX-based Transfer Function
Model Identification Using Wide-area Measurement for Adaptive and
Coordinated Damping Control”, IEEE Trans. Smart Grid, 2017, 1105-
1115.
[29]D. Bian, Z. Yu,D. Shi, R. Diao, Z. Wang, "A Robust Low-Frequency
Oscillation Detection and Analysis (LFODA) System with Innovative
Ensemble Filtering for Real-time Grid Operation," CSEE Journal of
Power and Energy Systems, accepted, 2019.
[30]X. Deng, G. He, Y. Chen, W. Zhang,“CIM Leadin Based on Java
Reflection Mechanism in AEMS of Shanghai Power Grid,” Automation
of Electric Power System, 2007, vol.31, no.18, pp. 21-24
[31]Y. Liu et al., "Recent developments of FNET/GridEye — A situational
awareness tool for smart grid," in CSEE Journal of Power and Energy
Systems, vol. 2, no. 3, pp. 19-27, Sept. 2016.
[32]H. Liu, J. Guo, W. Yu, T. Xia, R. Sun, M. Gardner, L. Zhu, Y.ilu Liu.
“The Design and Implementation of the Enterprise Level Data Platform
and Big Data Driven Applications and Analytics”, 2016 IEEE PES
Transmission & Distribution Conference & Exposition, 2015.
[33]X. Lu, D. Shi and et al., "PMU assisted power system parameter
calibration at Jiangsu electric power company," IEEE Power & Energy
Society General Meeting, Chicago, IL, 2017, pp. 1-5.
[34]D. Zhou, Y. Liu and J. Dong, "Frequency-based real-time line trip
detection and alarm trigger development," IEEE PES General Meeting,
National Harbor, MD, 2014, pp. 1-5.
[35]C. Li, Z. Wang and H. Qi, "Fast-Converging Conditional Generative
Adversarial Networks for Image Synthesis," 2018 25th IEEE
International Conference on Image Processing (ICIP), Athens, 2018, pp.
2132-2136.
[36]J. Zhao, L. Zhan, Y. Liu, H. Qi, J. R. Garcia, P. D. Ewing, “Measurement
accuracy limitation analysis on synchrophasors”, 2015 IEEE PES General
Meeting, Denver, CO, 2015.
[37]W. Yao, Y. Liu, D Zhou, Z Pan, J. Zhao, M. Till, L. Zhu, L. Zhan, Q.
Tang and Y. Liu, Impact of GPS Signal Loss and Its Mitigation in Power