A Hierarchical Approach to Dynamic Big Data Analysis in Power Infrastructure Security (NSF/AFOSR DDS EAGER Project 1462530) PIs: Hamed Mohsenian-Rad (EE), Christian Shelton (CS), Fabio Pasqualetti (ME)
University of California at Riverside
PI Meeting – Arlington, VA, January 2016
Problem Statement • Goal: Detect and identify faults and attacks from diverse multi-resolution dynamic data in a power infrastruture. Smart Meter (Consumer)
Market (Transmission)
8000
200
(Res: 1 sec)
7000
> 5000 Nodes
(Res: 5 min)
6000
150 $ / MWh
Watts
5000 4000 3000
100
50
2000 1000 0
0
6
Smart Meter
6
7
8
0
9
0
500
4
x 10
1000 Samples
1500
2000
Solar Panel (Consumer)
400
300
200 kW
MW / MVAR
4 5 Second
4 3
100
Q
2
0
(Res: 1 min)
1 0
Hamed Mohsenian-Rad
3
P
5
Feeder 1224 (12 kV)
2
Substation (Distribution)
7
Hunter Substation (69 kV)
1
0
2000
4000 6000 Samples
8000
(Res: ~1 min) 10000
-100
Dynamic Big Data Analysis in Power Infrastructure
0
1000
2000
3000 4000 Samples
5000
6000
7000
UC Riverside
1/9
Problem Statement • Goal: Detect and identify faults and attacks from diverse multi-resolution dynamic data in a power infrastruture. Swing Equations
States Frequency Voltage Magnitude Voltage Phase Angle Power Flow
Power Flow Equations
…
𝑥 = 𝑓 𝑥, 𝑢 𝑦 = 𝑔(𝑥, 𝑢)
Hamed Mohsenian-Rad
Dynamic Big Data Analysis in Power Infrastructure
UC Riverside
2/9
Problem Statement • Goal: Detect and identify faults and attacks from diverse multi-resolution dynamic data in a power infrastruture.
Generation Level Randomness
Load Level
Control
Randomness Control
𝑥 = 𝑓 𝑥, 𝑢 𝑦 = 𝑔(𝑥, 𝑢)
Hamed Mohsenian-Rad
Dynamic Big Data Analysis in Power Infrastructure
UC Riverside
3/9
Problem Statement • Goal: Detect and identify faults and attacks from diverse multi-resolution dynamic data in a power infrastruture. Positive Feedback Generation Level 𝑗𝜔 x
x
Fault
x
System Poles
x
Linear Model
Attack [DeMarco 1996, Pasqualetti 2012, etc.]
Hamed Mohsenian-Rad
Dynamic Big Data Analysis in Power Infrastructure
UC Riverside
4/9
Problem Statement • Goal: Detect and identify faults and attacks from diverse multi-resolution dynamic data in a power infrastruture.
(Distributed?) Positive Feedback Load Level 𝑓 Frequency Sensor
𝑓
Fault Attack [Mohsenian-Rad 2010, Marnerides 2014, Amini 2015, etc.]
Hamed Mohsenian-Rad
Dynamic Big Data Analysis in Power Infrastructure
UC Riverside
5/9
Research Challenges • Problem 1: Detection
Monitor 1
Interconnected System
Monitor 2
Hamed Mohsenian-Rad
Dynamic Big Data Analysis in Power Infrastructure
UC Riverside
6/9
Research Challenges • Problem 2: Identification (Recall > 5000 nodes, etc.) Location 1
Machine Learning Control Theory Power Systems
Time Domain Frequency Domain
Location 2 FFT Magnitude
0.08 0.06 0.04 0.02 0 0
Hamed Mohsenian-Rad
Dynamic Big Data Analysis in Power Infrastructure
Attack Signature
0.1
0.2 0.3 0.4 Frequency (Hz)
UC Riverside
0.5
7/9
Project Planning • Parallel Research Efforts: • Task 1: Problem Formulation
• Task 2: Designing Detection / Monitoring Tools • Task 3: Designing Identification Tools • Additional Task: – Developing Proper [Hierarchical] Simulation and Testing Tools
Hamed Mohsenian-Rad
Dynamic Big Data Analysis in Power Infrastructure
UC Riverside
8/9
Research Plan
Thank You! E-mail:
[email protected] Web: www.ece.ucr.edu/~hamed
Hamed Mohsenian-Rad
Dynamic Big Data Analysis in Power Infrastructure
UC Riverside
9/9
Research Plan [1] C. L. DeMarco, J. V. Sariashkar, and F. Alvarado, “The potential for malicious control in a competitive power systems environment,” in Proc. of the IEEE International Conference on Control Applications, Dearborn, MI, 1996. [2] F. Pasqualetti and F. Dörfler and F. Bullo “Cyber-Physical Security via Geometric Control: Distributed Monitoring and Malicious Attacks” in Proc. of the IEEE Conference on Decision and Control, Maui, Hi, USA, 2012. [3] A. H. Mohsenian-Rad and A. Leon-Garcia, “Distributed Internet-based load altering attacks against smart power grids”, IEEE Trans. Smart Grid, vol. 2, no. 4, pp.667-674, December 2011. [4] A. K. Marnerides, P. Smith, P. A. Schaeffer-Filho, and A. Mauthe, “Power Consumption Profiling Using Energy Time-Frequency Distributions in Smart Grids”, IEEE Communications Letters, vol. 19, no. 1, pp. 46-49, January 2015.
[5] S. Amini and F. Pasqualetti and H. Mohsenian-Rad, “Detecting Dynamic Load Altering Attacks: A Data-Driven Time-Frequency Analysis” in Proc. of the IEEE International Conference on Smart Grid Communications, Miami, FL, 2015. [6] S. Amini and H. Mohsenian-Rad and F. Pasqualetti, "Dynamic Load Altering Attacks in Smart Grid", in Proc. of the IEEE PES Conference on Innovative Smart Grid Technologies, Washington, DC, 2015.
Hamed Mohsenian-Rad
Dynamic Big Data Analysis in Power Infrastructure
UC Riverside
R