North Carolina State University Undergraduate Symposium





2012- 21st Annual NC State Undergraduate Research Symposium

Close Details
 
Session Time : 4/10/12 10:30 AM - 4/10/12 11:45 AM
Content Area : Electrical & Computer Engineering
Poster Appointment: NONE
Student Presenters :       
Joel E Anderson Computer Science
Mentors and/or Co-Authors :
Aranya Chakrabortty Elec & Comp Engineering
Abstract Title : Graph-Theoretic Algorithms for Optimal Sensor Placement and Malicious Attack Detection in Large Power Grids
Abstract :
The time-varying stochastic nature of power system operations require constant monitoring of its physical variables to avoid catastrophic failures. Modern measurement technologies such as Wide-Area Measurement Systems (WAMS) allow us to track these variables in real-time using very high-resolution dynamic data available from Phasor Measurement Units (PMUs). Due to cost constraints, however, system operators typically need the total number of PMUs to be installed in the system to be minimal, under which situation one must determine the optimal locations of these PMUs guaranteeing complete observability of the network. While algorithms for such minimal sensor placement are well-known in graph theory for an unconstrained measurement capacity of the PMUs, deriving them for PMUs with channel constraints still remains a challenge. Motivated by this problem, in Part-I of the project we develop a sub-optimal minimum cover algorithm for PMU placement under such measurement constraints, and illustrate their sensitivity to network size and complexity. In Part-II we develop a numerical method to track if any of the PMUs in the derived minimum cover set might be compromised by external malicious attacks. By placing the PMU data into a special matrix, we validate the measurement values within a specific tolerance using a symmetry-preserving algorithm, and tune it to balance accuracy of the solutions with computational costs. Both algorithms are illustrated via realistic power system examples, and implemented in our recently developed power system visualization software platform Watchdog.