Informs Annual Meeting 2017

WA73

INFORMS Houston – 2017

WA74

4 - On the Structure of the Inverse-feasible Region of a Linear Program

372C Joint session ICS/ Practice: PMU based Data Analytics and Machine Learning in Power Grids Sponsored: Computing Sponsored Session Chair: Deepjyoti Deka, Los Alamos National Lab, Los Alamos, NM, 87544, United States, deepjyotideka@gmail.com Co-Chair: Andrey Lokhov, Los Alamos National Laboratory, CNLS, MS B258, Los Alamos, NM, 87545, United States, lokhov@lanl.gov 1 - Real-time Data-driven Detection of Low-quality Synchrophasor Measurements Meng Wu, PhD Student, Texas A&M.University, College Station, TX, 77840, United States, mariewumeng@gmail.com, Le Xie An online data-driven approach is proposed for the detection of low-quality synchrophasor measurements. The proposed method leverages the spatio- temporal similarities among multiple-time-instant synchrophasor data, and formulates the low-quality synchrophasor data as spatio-temporal outliers. A density-based local outlier detection technique is proposed to detect the spatio- temporal outliers. This data-driven approach involves no system modeling information. The algorithm can operate under both normal and fault-on system conditions, with fast computation speed suitable for online applications. The effectiveness of the proposed approach is verified via case studies. 2 - Smart Grid Security through Synchrophasor Data: Real-time Detection of Attacks on AC State Estimation Biplab Sikdar, PhD, National University of Singapore, bsikdar@nus.edu.sg One of the most important functions in the control and monitoring of a power grid is AC state estimation whose output is used by Energy Management System (EMS) applications such as economic dispatch, contingency analysis, and optimal power flow. Protecting the state estimator against cyber-attacks that bias its outputs is thus of critical importance. This paper proposes a simple non-iterative technique for detecting false data injection attacks on AC state estimators, by using synchrophasor data. As the proposed method is independent of the state estimator outputs and does not depend on any EMS functionality, it can be used to test the data even before the execution of the state estimation algorithm. 3 - Reconstructing the Power Grid Dynamic Model from Sparse Measurements Andrey Y. Lokhov, Los Alamos National Laboratory, CNLS, MS.B258, Los Alamos, NM, 87545, United States, lokhov@lanl.gov, Deepjyoti Deka, Marc Vuffray, Michael Chertkov We describe the problem of parameter reconstruction in the power grid dynamics from sparsely located phasor measurement units (PMUs) observations. We assume that the dynamics is described by a system of coupled linearized second- order differential swing equations. In the ambient regime, fluctuations on individual nodes are assumed to be independent and Gaussian distributed. The goal is to reconstruct the inertia and damping coefficients of generators, as well as line susceptances. We discuss how this learning task can be efficiently combined with the problem of state estimation of consumption in the grid, and solved using regularized maximum likelihood and least-squares estimators. 4 - High-dimensional Data Analytics of PMU Measurements by Exploiting Low-dimensional Models Meng Wang, Rensselaer Polytechnic Institute, Troy, NY, United States, wangm7@rpi.edu The increasingly denser coverage of phasor measurement units (PMUs) enables dynamic visibility into power systems. The large amounts of data obtained by PMUs impose significant challenges to data management and information extraction. Interestingly, the spatial-temporal blocks of PMU data exhibit low- dimensional structures despite the high ambient dimensions. Such low-dimensionality can be exploited to enable and simplify multiple PMU data analysis tasks under the same framework.

Onur Tavaslioglu, University of Pittsburgh, 7171 Buffalo Speedway, Apt 1934, Houston, TX, 77025, United States, ont1@pitt.edu, Taewoo Lee, Andrew J.Schaefer

Given a feasible solution, inverse linear programming considers perturbing the objective vector of a linear program such that the solution is optimal and the perturbation is minimum under an Lp norm. We study the polyhedral characterization of the feasible region of an inverse linear programming problem. We characterize the inverse-feasible region to show the relationship between the dimensions of the minimum faces and inverse-feasible regions, which gives necessary and sufficient conditions of the extreme, boundary, and inner points of a linear program. We characterize the set of objective vectors that generate a selected unique optimal solution for a considered linear program.

WA73

372B Power System Reliability Invited: Energy Systems Management Invited Session

Chair: Mohammad Najarian, University of Houston, 4722 Calhoun Road, Engineering Building 2, Houston, TX, 77204-4008, United States, mohammad.najarian@gmail.com 1 - Designing Resilient Electric Networks under Natural Hazards Tomas Ignacio Lagos, University of Chile, Beauchef 851, Santiago, 7640031, Chile, tomas.lagos.gonzalez@gmail.com We propose an optimization framework for designing a resilient electric power grid under high impact low probability events such as earthquakes. An Optimization via Simulation approach is used to solve this discrete decision problem, where the measure of resilience is the expected energy not supplied, assessed through an existing simulator that contains historical earthquake data, fragility curves of the network components, and a unit commitment model. We use this framework to discuss various approximation methods applied to select subsets of scenarios and take optimal decisions on network investment. 2 - Modeling Dependent Outages of Electricity Power Plants Vishwakant Malladi, University of Texas-Austin, McCombs School of Business, CBA 5.202, IROM, Austin, TX, 78712, United States, vishwakant@gmail.com, Stathis Tompaidis, Rafael Mendoza We propose a new framework to model dependence of outages of electricity power plants. Our framework accounts for commonality in outages, such as weather events and fuel shortages using a unique data set of actual outages. We calibrate our model for the Electricity Reliability Council of Texas (ERCOT) and the Western Electricity Coordinating Council (WECC) regions. We find strong evidence of dependence in outages both in the input fuel and the geographic location of the power plants and discuss the consequences of this dependence on the reliability of electricity supply. 3 - Resilience Assessment of Interdependent Gas Network and Electrical Power System Infrastructures: A Quantitative Approach Mohammad Najarian, PhD Student, University of Houston, E206, Engineering Bldg 2, 4722 Calhoun Rd, Houston, TX, 77204-4008, United States, mohammad.najarian@gmail.com, Masoud Barati, Gino J. Lim To reduce the consequences of hazards resilience must be incorporated into the design of the electrical power systems(EPS) and it must be quantified.The interdependence between EPS and other critical infrastructures, such as natural gas systems (NGS), compounds the complexity of resilience measurement. In this paper, we provide a method to measure and analyze the impact of NGS resilience on the EPS resilience. This method is also used to assign limited resources to both EPS and NGS aiming at increasing the resilience of the EPS. For this, a combination of security-constrained unit commitment with gas dependency constraint and Monte Carlo simulation is used.

468