Professional Documents
Culture Documents
Enhancing Smart Grid Security and Reliability Through Graph Signa
Enhancing Smart Grid Security and Reliability Through Graph Signa
USF Tampa Graduate Theses and Dissertations USF Graduate Theses and Dissertations
June 2023
This Dissertation is brought to you for free and open access by the USF Graduate Theses and Dissertations at
Digital Commons @ University of South Florida. It has been accepted for inclusion in USF Tampa Graduate Theses
and Dissertations by an authorized administrator of Digital Commons @ University of South Florida. For more
information, please contact digitalcommons@usf.edu.
List of Tables ................................................................................................................. vi
Abstract ......................................................................................................................... xi
ii
3.4.2 Feature Extraction and Classification for Stress Detection ....... 62
3.2.2.1 k-NN Method for Classification ..................................... 64
3.4.3 Locating Stresses in the System .................................................. 65
3.4.4 Computational Complexity .......................................................... 66
3.4.5 Simulation and Results ................................................................ 67
3.4.5.1 Generating State Attribute Time Series ....................... 67
3.4.5.2 Detection and Locating Performance ............................. 68
iii
Detection, ............................................................................ 96
5.3.1.2 Features Extracted Using GFT of Temporal
Statistics ................................................................................. 97
5.3.1.3 Features Extracted by Taking Temporal
Statistics of the Time-Varying GFT Values .......................... 98
5.3.2 Dimensionality Reduction of the GFT-based Features .............. 98
5.4 Simulation Details ..................................................................................... 99
5.4.1 Two-stage Classification .............................................................. 99
5.4.2 Characterization ........................................................................... 99
5.5 Performance Evaluation .......................................................................... 100
5.5.1 Performance of the Two-stage Classification ............................ 100
5.5.1.1 Noise Sensitivity of Classification Performance ......... 102
5.5.1.2 Classification Performance with a Reduced
Number of Features .............................................................. 103
5.5.2 Classification Performance for Multiple Random Vs.
Clustered Cyber Attacks................................................................. 104
5.5.3 Determining and in Clustered Cyber Attacks .................. 105
iv
6.5.1 System and Attack Models......................................................... 124
6.5.1.1 Power System Model .................................................... 124
6.5.1.2 DoS Attack Model ......................................................... 124
6.5.2 Method ........................................................................................ 125
6.5.3 Simulations and Results ............................................................ 127
6.5.3.1 Correlation Among PMU Time Series ......................... 128
6.5.3.2 Estimation of PMU Time Series Under Single
and Multiple DoS Attacks .................................................... 128
6.5.3.3 Most Vulnerable Combination of Attacks.................... 131
6.5.3.4 Parameter Values ..................................................... 132
6.5.3.5 Most Vulnerable Time for the DoS Attack .................. 132
v
Replay
Attack
FDIA Delay
DoS Attack
Attack
Ramp
Attack
Error caused by the Anti-Aliaing Filter
Error caused by the Reconstruction using Matrix Operations
Total Error of Sampling-Reconstruction
Mean squared Error
FDIA Delay
DoS Attack
Attack
Ramp
Attack
1
0.8
Accuracy
0.6
0.4
0.95
Accuracy 0.9
0.85
Random Vs. Clustered Classification
Attack Center Locating (Clustered)
Attack Radius Estimation (Clustered)
0.8
1 2 3 4 5
Amount of delay (samples)
Mean Absolute Error of Recovery (in dB)
Clustered Cyber-Attack
Random Cyber-Attack
0.8
0.6
0.4
0.2
0
5 10 15 20
Number of Unobservable Buses
Random
PMU setting
31
30.5
30
0 5 10 15 20 25
Time (Hours)
-12
Original Voltage Angle
Estimated Voltage Angle
The DoS Attack Moment
-12.5
Voltage Angle (in Degree)
-13
-13.5
-14
-14.5
0 5 10 15 20 25
Time (Hours)
0.99
Original Voltage Magnitude
Estimated Voltage Magnitude
The DoS Attack Moment
0.989
0.9885
0.988
0 5 10 15 20 25
Time (Hours)
31.5
31
30.5
30
0 5 10 15 20 25
Time (Hours)
0.25
Uniform
Clustered
Inhibition
0.2
0.1
0.05
0
0 20 40 60 80 100 120
Number of Attackeded PMUs
29.5 a=1
a=5
a=100
29 a=200
Voltage Angle (in Degree)
a=500
a=700
28.5 Original Bus Voltage Angle
The Moment of DoS Attack
28
27.5
27
26.5
0 5 10 15 20 25
Time (Hours)
0.1
Day 1
Day 2
0.08
Mean squared error
0.06
0.04
0.02
0
0 5 10 15 20 25
Time of the day (Hour)