Situational Awareness Using Distribution Synchrophasors ... · Hamed Mohsenian-Rad Situational...
Transcript of Situational Awareness Using Distribution Synchrophasors ... · Hamed Mohsenian-Rad Situational...
Situational Awareness Using Distribution Synchrophasors:
Application to Asset Monitoring
Hamed Mohsenian-Rad
Associate Professor, Electrical Engineering, University of California, RiversideAssociate Director, Winston Chung Global Energy Center
Director, UC-National Lab Center for Power Distribution Cyber Security
Talk at WCGEC ESTAP, 4/11/2019
Acknowledgements: A. Shahsavari, M. Farajollahi, E. Stewart, E. Cortez, A. Von-Meier, L. Alvarez, C. Roberts, F. Megala, Z. Taylor
Situational Awareness Using Distribution Synchrophasors:
Application to Asset Monitoring
Hamed Mohsenian-Rad
Associate Professor, Electrical Engineering, University of California, RiversideAssociate Director, Winston Chung Global Energy Center
Director, UC-National Lab Center for Power Distribution Cyber Security
Talk at WCGEC ESTAP, 4/11/2019
Acknowledgements: A. Shahsavari, M. Farajollahi, E. Stewart, E. Cortez, A. Von-Meier, L. Alvarez, C. Roberts, F. Megala, Z. Taylor
Distributed Energy Storage
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 1 / 19
Distributed Energy Storage Resources
Circuit 6
...
...
Substation: 69 kV / 12 kV
OLTC Fuse
Capacitor Bank
CircuitBreaker
Batteries
Distribution Feeder
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 1 / 19
Distributed Energy Storage Resources
Circuit 6
...
...
Substation: 69 kV / 12 kV
OLTC Fuse
Capacitor Bank
CircuitBreaker
Batteries
Distribution Feeder
• They are often not monitored directly.
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 1 / 19
Distributed Energy Storage Resources
Circuit 6
...
...
Substation: 69 kV / 12 kV
OLTC Fuse
Capacitor Bank
CircuitBreaker
Batteries
Distribution Feeder
• They are often not monitored directly.
• They are potential targets for cyber attacks physical botnet
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 1 / 19
Distributed Energy Storage Resources
Circuit 6
...
...
Substation: 69 kV / 12 kV
OLTC Fuse
Capacitor Bank
CircuitBreaker
Batteries
Distribution Feeder
• They are often not monitored directly.
• They are potential targets for cyber attacks physical botnet
• We want to monitor their Health and Security?
Remotely!
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 1 / 19
Distributed Energy Storage Resources
Circuit 6
...
...
Substation: 69 kV / 12 kV
OLTC Fuse
Capacitor Bank
CircuitBreaker
Batteries
Distribution Feeder
• They are often not monitored directly.
• They are potential targets for cyber attacks physical botnet
• We want to monitor their Health and Security?
Remotely!Micro-PMUs (GPS-equipped Phasor Sensors)
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 1 / 19
Distributed Energy Storage Resources
Circuit 6
...
...
Substation: 69 kV / 12 kV
OLTC Fuse
Capacitor Bank
CircuitBreaker
Batteries
Distribution Feeder
• They are often not monitored directly.
• They are potential targets for cyber attacks physical botnet
• We want to monitor their Health and Security?
Remotely & Automatically!Micro-PMUs (GPS-equipped Phasor Sensors)
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 2 / 19
Event-Driven Study
Sensor
Phase A
120 fps
10,368,000 measurement points
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 2 / 19
Event-Driven Study
Sensor
Phase A
120 fps
10,368,000 measurement points
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 2 / 19
Event-Driven Study
Sensor
120 fps
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 2 / 19
Event-Driven Study
Sensor
120 fps
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 3 / 19
Phase 1: Event Detection
Sensor
Window
Anomaly Detection
Absolute Deviation Around the Median
120 fps
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 3 / 19
Phase 1: Event Detection
Sensor
Window
Anomaly Detection
Absolute Deviation Around the Median
120 fps
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 3 / 19
Phase 1: Event Detection
Sensor
Anomaly Detection
Absolute Deviation Around the Median
Slide Window
120 fps
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 3 / 19
Phase 1: Event Detection
Sensor
Anomaly Detection
Absolute Deviation Around the Median
Event 1
120 fps
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 3 / 19
Phase 1: Event Detection
Sensor
Anomaly Detection
Absolute Deviation Around the Median
Event 1
Window
120 fps
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 3 / 19
Phase 1: Event Detection
Sensor
Anomaly Detection
Absolute Deviation Around the Median
Event 1 Event 2
120 fps
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 3 / 19
Phase 1: Event Detection
Sensor
Anomaly Detection
Absolute Deviation Around the Median
Event 1 Event 2
120 fps
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 3 / 19
Phase 1: Event Detection
Sensor
Anomaly Detection
Event 1 Event 2
120 fps
Current (𝐼)
Voltage (𝑉)
Active Power (𝑃)
Reactive Power (𝑄)
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 3 / 19
Phase 1: Event Detection
Sensor
Event 1 Event 2
120 fps
On Average: 500 Events Per Day Per Feeder
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 4 / 19
Phase 2: Event Classification
Micro-PMU 1 Micro-PMU 2
Class 1
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 4 / 19
Phase 2: Event Classification
Micro-PMU 1 Micro-PMU 2
Class 2
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 4 / 19
Phase 2: Event Classification
Micro-PMU 1 Micro-PMU 2Class 3
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 5 / 19
Phase 2: Event Classification
• Methodology:
Data Driven
Unsupervised Classification
Supervised Classification
Labeling Training Samples (Classes 1, 2, and 3)
Feature Extraction
Power Engineering
Model Based
Utility Records
2309 Events
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 6 / 19
Phase 2: Event Classification
• Classification Results:
Class 2: 296 1896
Class 1: 368 1434
Class 3: 1645 5018
Features
Detection Signal
Detection Window
Statistics
Difference
Correlation
min{𝑊}
𝑋𝑖 ∈ 𝐼, 𝑉, 𝑃, 𝑄
𝜎(𝑋𝑖)
𝑋𝑖𝑢 − 𝑋𝑖
𝑑
𝜌(𝑋𝑖 , 𝑌𝑖)
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 6 / 19
Phase 2: Event Classification
• Classification Results:
Class 2: 296 1896
Class 1: 368 1434
Class 3: 1645 5018
Features
Detection Signal
Detection Window
Statistics
Difference
Correlation
min{𝑊}
𝑋𝑖 ∈ 𝐼, 𝑉, 𝑃, 𝑄
𝜎(𝑋𝑖)
𝑋𝑖𝑢 − 𝑋𝑖
𝑑
𝜌(𝑋𝑖 , 𝑌𝑖)
Training Data
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 6 / 19
Phase 2: Event Classification
• Classification Results:
Class 2: 296 1896
Class 1: 368 1434
Class 3: 1645 5018
Features
Detection Signal
Detection Window
Statistics
Difference
Correlation
min{𝑊}
𝑋𝑖 ∈ 𝐼, 𝑉, 𝑃, 𝑄
𝜎(𝑋𝑖)
𝑋𝑖𝑢 − 𝑋𝑖
𝑑
𝜌(𝑋𝑖 , 𝑌𝑖)
Test Data
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 6 / 19
Phase 2: Event Classification
• Classification Results:
Class 2: 296 1896
Class 1: 368 1434
Class 3: 1645 5018
Features
Detection Signal
Detection Window
Statistics
Difference
Correlation
min{𝑊}
𝑋𝑖 ∈ 𝐼, 𝑉, 𝑃, 𝑄
𝜎(𝑋𝑖)
𝑋𝑖𝑢 − 𝑋𝑖
𝑑
𝜌(𝑋𝑖 , 𝑌𝑖)
Tra
inin
g D
ata
Te
st
Da
ta
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 8 / 19
Phase 3: Event Source Identification
Distribution Substation
OtherFeeders
…
1 2
3-5
8-1
2
6 7 13
14-19
20 28 29 30
21
-27
𝐼𝑢 ∠𝐼𝑢𝑉𝑢 ∠𝑉𝑢,
Questions: What is the source location of this event?
𝐼𝑑 ∠𝐼𝑑𝑉𝑑 ∠𝑉𝑑,
Upstream Downstream
Class 3
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 9 / 19
Phase 3: Event Source Identification
_+ V
_+
Vpre
Ipre Z
pre
_+ V
post
Ipost Z
post
Vposts V
postr
Ipostsr
Vpre
s Vpre
rI
presr
Vs Vr Isr
I
Pre-Event
Event
Post-Event
V_
+
Vr Isr
I
Equivalent Circuit
∆𝑉 = 𝑉𝑝𝑜𝑠𝑡 − 𝑉𝑝𝑟𝑒
∆𝐼 = 𝐼𝑝𝑜𝑠𝑡 − 𝐼𝑝𝑟𝑒
Sou
rces
Rem
ove
d
Theoretical Foundation: Compensation Theorem
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 10 / 19
Step 3-1: Extract Differential Synchrophasors
Upstream Sensor Downstream Sensor
𝐼𝑢∠𝐼𝑢
𝑉𝑢
∠𝑉𝑢
𝐼𝑑∠𝐼𝑑
𝑉𝑑
∠𝑉𝑑
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 10 / 19
Step 3-1: Extract Differential Synchrophasors
Upstream Sensor Downstream Sensor
𝐼𝑢∠𝐼𝑢
𝑉𝑢
∠𝑉𝑢
𝐼𝑑∠𝐼𝑑
𝑉𝑑
∠𝑉𝑑
∆𝐼𝑢 ∠∆𝐼𝑢
Pre
-Eve
nt
Post
-Eve
nt
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 10 / 19
Step 3-1: Extract Differential Synchrophasors
Upstream Sensor Downstream Sensor
𝐼𝑢∠𝐼𝑢
𝑉𝑢
∠𝑉𝑢
𝐼𝑑∠𝐼𝑑
𝑉𝑑
∠𝑉𝑑
∆𝐼𝑢 ∠∆𝐼𝑢
Pre
-Eve
nt
Post
-Eve
nt
∆𝑉𝑢 ∠∆𝑉𝑢
Pre
-Eve
nt
Post
-Eve
nt
∆𝐼𝑑 ∠∆𝐼𝑑P
re-E
ven
t
Post
-Eve
nt
∆𝑉𝑑 ∠∆𝑉𝑑
Pre
-Eve
nt
Post
-Eve
nt
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 11 / 19
Step 3-2: Forward Nodal Voltage Calculation
Y1
Z1 Zn-1
Ik
Vn Vk
|Vd| V
d
|Id| I
dZ
u Zd
V2 Vn-1 V1
I1
Y2
I2
Yk
Ik
Yn-1
In-1
Yn
In
Downstream
|Vu| V
u
|Iu| I
u
Upstream
Zk-1 Vk-1
Ik-1
Zk Vk+1
Ik+1
𝑘: Event Bus (Unknown)Forward
∆𝑉1𝑓= ∆𝑉𝑢
∆𝑉2𝑓= ∆𝑉1
𝑓+ ∆𝐼𝑢 + 𝑌1∆𝑉1
𝑓𝑍1
⋮
∆𝑉𝑘𝑓= ∆𝑉𝑘−1
𝑓+ ∆𝐼𝑢 + 𝑌1∆𝑉1
𝑓+⋯+ 𝑌𝑘−1∆𝑉𝑘−1
𝑓𝑍𝑘−1
∆𝑉𝑘+1𝑓
≠ ∆𝑉𝑘𝑓+ ∆𝐼𝑢 + 𝑌1∆𝑉1
𝑓+⋯+ 𝑌𝑘−1∆𝑉𝑘−1
𝑓+ 𝑌𝑘∆𝑉𝑘
𝑓𝑍𝑘
I
∆𝑉𝑛𝑓≠ ∆𝑉𝑛−1
𝑓+ ∆𝐼𝑢 + 𝑌1∆𝑉1
𝑓+⋯+ 𝑌𝑛−1∆𝑉𝑛−1
𝑓𝑍𝑛−1
⋮
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 12 / 19
Step 3-3: Backward Nodal Voltage Calculation
Y1
Z1 Zn-1
Ik
Vn Vk
|Vd| V
d
|Id| I
dZ
u Zd
V2 Vn-1 V1
I1
Y2
I2
Yk
Ik
Yn-1
In-1
Yn
In
Downstream
|Vu| V
u
|Iu| I
u
Upstream
Zk-1 Vk-1
Ik-1
Zk Vk+1
Ik+1
𝑘: Event Bus (Unknown) Backward
II
∆𝑉𝑛−1𝑏 = ∆𝑉𝑛
𝑏 + ∆𝐼𝑑 + 𝑌𝑛∆𝑉𝑛𝑏 𝑍𝑛−1
⋮
∆𝑉𝑛𝑏 = ∆𝑉𝑑
∆𝑉𝑘𝑏 = ∆𝑉𝑘+1
𝑏 + ∆𝐼𝑢 + 𝑌𝑛∆𝑉𝑛𝑏 +⋯+ 𝑌𝑘+1∆𝑉𝑘+1
𝑏 𝑍𝑘
∆𝑉𝑘−1𝑏 ≠ ∆𝑉𝑘
𝑏 + ∆𝐼𝑢 + 𝑌𝑛∆𝑉𝑛𝑏 +⋯+ 𝑌𝑘+1∆𝑉𝑘+1
𝑏 + 𝑌𝑘∆𝑉𝑘𝑏 𝑍𝑘−1
∆𝑉1𝑏 ≠ ∆𝑉2
𝑏 + ∆𝐼𝑢 + 𝑌𝑛∆𝑉𝑛𝑏 +⋯+ 𝑌2∆𝑉2
𝑏 𝑍1
⋮
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 13 / 19
Step 3-4: Analysis of Discrepancy Index
Y1
Z1 Zn-1
Ik
Vn Vk
|Vd| V
d
|Id| I
dZ
u Zd
V2 Vn-1 V1
I1
Y2
I2
Yk
Ik
Yn-1
In-1
Yn
In
Downstream
|Vu| V
u
|Iu| I
u
Upstream
Zk-1 Vk-1
Ik-1
Zk Vk+1
Ik+1
𝑘: Event Bus (Unknown) BackwardForward
∆𝑉1𝑓, ⋯ , ∆𝑉𝑘−1
𝑓, ∆𝑉𝑘
𝑓, ∆𝑉𝑘+1
𝑓, ⋯ , ∆𝑉𝑛
𝑓
∆𝑉1𝑏 , ⋯ , ∆𝑉𝑘−1
𝑏 , ∆𝑉𝑘𝑏, ∆𝑉𝑘+1
𝑏 , ⋯ , ∆𝑉𝑛𝑏
Correct
Correct
Incorrect
Incorrect
Forward:
Backward:
𝑘 = argmin ∆𝑉𝑖𝑓− ∆𝑉𝑖
𝑏
𝑖
Φ𝑖
Discrepancy Index
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 14 / 19
Phase 3: Event Source Identification
31 2 4 5
22 3031
32
33
34
3536
37 38 39
41
40
42
4344
6 7 8 9
1011
12 13
14
15
1617181920
21
23
24
25
26
27 28 29
Mic
ro-P
MU
1
Micro-PMU 2
Mic
ro-P
MU
4
Micro-PMU 3
Event Location
Event Location
Event Location
Discrepancy Index
IEEE
12
3-B
us
Test
Sys
tem
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 14 / 19
Phase 3: Event Source Identification
31 2 4 5
22 3031
32
33
34
3536
37 38 39
41
40
42
4344
6 7 8 9
1011
12 13
14
15
1617181920
21
23
24
25
26
27 28 29
Mic
ro-P
MU
1
Micro-PMU 2
Mic
ro-P
MU
4
Micro-PMU 3
Event Location
Event Location
Event Location
Discrepancy Index
IEEE
12
3-B
us
Test
Sys
tem
Event Source Location
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 14 / 19
Phase 3: Event Source Identification
31 2 4 5
22 3031
32
33
34
3536
37 38 39
41
40
42
4344
6 7 8 9
1011
12 13
14
15
1617181920
21
23
24
25
26
27 28 29
Mic
ro-P
MU
1
Micro-PMU 2
Mic
ro-P
MU
4
Micro-PMU 3
Event Location
Event Location
Event Location
Discrepancy
IEEE
12
3-B
us
Test
Sys
tem
Event Source Location
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 14 / 19
Phase 3: Event Source Identification
31 2 4 5
22 3031
32
33
34
3536
37 38 39
41
40
42
4344
6 7 8 9
1011
12 13
14
15
1617181920
21
23
24
25
26
27 28 29
Mic
ro-P
MU
1
Micro-PMU 2
Mic
ro-P
MU
4
Micro-PMU 3
Event Location
Event Location
Event Location
Discrepancy
IEEE
12
3-B
us
Test
Sys
tem
Event Source Location
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 15 / 19
• Data-Analytics Package for Distribution Synchrophasors
• Event Detection
• Event Classification
• Event Source Identification
• Case Study: Remote Asset Monitoring
Recap: Big Data?
Billions of Data Points
Device-Specific Event Signatures
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 15 / 19
• Data-Analytics Package for Distribution Synchrophasors
• Event Detection
• Event Classification
• Event Source Identification
• Case Study: Remote Asset Monitoring
Recap: Big Data?
Billions of Data Points
Device-Specific Event Signatures
Distributed Energy Storage
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 16 / 19
Case Study: Asset Monitoring
Three-Phase Switched Capacitor Bank
Rating: 3 x 300 kVAR = 900 kVAR
Volt/VAR Control
Onsite Switch On / Switch Off Controller
No Monitoring
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 16 / 19
Case Study: Asset Monitoring
Typical Issues:
Unbalanced Operation (Fuses) Switching Operation (Controllers)1 2
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 16 / 19
Case Study: (Remote) Asset Monitoring
Typical Issues:
Unbalanced Operation (Fuses) Switching Operation (Controllers)1 2
Remote Monitoring?
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 17 / 19
Case Study: (Remote) Asset Monitoring
Switch Off Event
(Micro-PMU 1)
Detection & Classification
Location Identification
(Micro-PMU 2)
Event Bus: 25 (Correct)
Discrepancy Index
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 17 / 19
Case Study: (Remote) Asset Monitoring
Switch Off Event
(Micro-PMU 1)
Detection & Classification
Reactive Power Support
Slightly Unbalanced Operation
Phase B is always higher
Likely fuse blowing on C and A
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 17 / 19
Case Study: (Remote) Asset Monitoring
Switch Off Event
(Micro-PMU 1)
Detection & Classification
Switching Transient
Two-Step 3-Phase Switch
Step 1: Phase C (Zero Crossing)
Step 2: Phase A/B (Possible Malfunction)
C
B
A
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 17 / 19
Case Study: (Remote) Asset Monitoring
Switch Off Event
(Micro-PMU 1)
Detection & Classification
Switching Transient
Two-Step 3-Phase Switch
Step 1: Phase C (Zero Crossing)
Step 2: Phase A/B (Possible Malfunction)
C ⟼ A/B
UnbalancedSystem
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 18 / 19
Further Reading
IEEE PES Magazine 2018 IEEE TPWRS 2018 IEEE TSG 2019
Hamed Mohsenian-Rad Situational Awareness Using Distribution Synchrophasors UC Riverside 19 / 19
Further Discussion
Alireza Shahsavari Mohammad Farajollahi