2016 INFORMS Annual Meeting Program

WD07

INFORMS Nashville – 2016

WD04 101D-MCC Robust and Stochastic Optimization for Energy Systems Sponsored: Energy, Natural Res & the Environment, Energy I Electricity Sponsored Session

points that causes the maximal damage. We then investigate the relationship between fire response capacity and the rate of spread, fire ignition location and number of fire ignitions in the landscape. 2 - An Optimization Model For Wildfire Suppression Andres L Medaglia, Professor and Chair, Universidad de los Andes, Cr 1 este #19 A 40, ML708, Bogota, Cundinamarca, 111711, Colombia, amedagli@uniandes.edu.co An effective early attack is essential to control wildfires. In this work, we propose an MIP to support decisions related to the planning and response phases of fire management. For the planning phase, the MIP addresses the decisions related to the location of facilities and how much inventory to store. In the response phase, decisions are concerned to the location of coordination centers and how to allocate available resources. The model includes a risk measure to limit the downside risk of different scenarios. We apply this methodology in a setting that resembles wildfires nearby the city of Bogotá (Colombia). WD06 102A-MCC Optimization in Data Mining 1 Sponsored: Data Mining Sponsored Session Chair: Orestis Panagopoulos, 9303 Nelson Park Circle, Apt 204, Orlando, FL, 32817, United States, ore.pan@hotmail.com 1 - New Developments Of L1 Splines: Fast Computation And Shape-preserving Capability Ziteng Wang, Assistant Professor, Northern Illinois University, DeKalb, IL, United States, th2168@gmail.com Cubic splines are widely used for data interpolation and approximation in terrain surface fitting, computer aided design, and numerical control. Conventional L2- norm based splines often show undesired oscillation and do not preserve shape, especially for irregular or multiscale data. L1 splines, by minimization of the L1- norm based metrics, have shown superior and robust shape-preserving performances and enjoyed increasing application potentials. We introduce the development of L1 splines over the past decade, present the latest research on the fast computing strategy and the quantitative measure of shape-preserving capability, and discuss future opportunities. 2 - Probing The Pareto Frontier Of Computational Statistical Tradeoffs Zhaoran Wang, Princeton University, Princeton, NJ, 08540, United States, zhaoran@princeton.edu In this talk, we discuss the fundamental tradeoffs between computational efficiency and statistical accuracy in big data. Based on an oracle computational model, we introduce a systematic hypothesis-free approach for developing minimax lower bounds under computational budget constraints. This approach mirrors the classical Le Cam’s method, and draws explicit connections between algorithmic complexity and geometric structures of parameter spaces. Based on this approach, we sharply characterize the computational-statistical phase transitions that arise in a broad class of learning problems. WD07 102B-MCC Predictive Analysis and Applications in Data Mining Sponsored: Data Mining Sponsored Session Chair: Juxihong Julaiti, Penn State University, 445 Waupelani Drive, Apt J1, State College, PA, 16801-4445, United States, juxihongjulaiti1225@gmail.com 1 - Proactive Data: A Rich Source Of Occupational Accident Prediction Jhareswar Maiti, Professor, IIT, Kharagpur, Kharagpur, 721302, India, jhareswar.maiti@gmail.com, Sobhan Sarkar, Saicharan Pardhu, Rutwick Ayi The proactive data ( e.g., inspection reports) is a source of rich information for prediction of occurrence of accidents. The main aim of the study is to use the proactive data properly along with reactive data (e.g., incident reports) retrieved from an integrated steel industry in building a prediction model to predict the occurrences of future incidents. Decision tree algorithms like C5.0, CART, CHAID, exhaustive CHAID, and ensemble techniques i.e., boosting have been implemented in order to predict the accidents. Results show that the C5.0 outperforms all other algorithms in terms of higher accuracy.

Chair: Andy Sun, Georgia Institute of Technology, 755 Ferst Dr, Atlanta, GA, 30312, United States, andy.sun@isye.gatech.edu 1 - Robust Optimization For The Alternating Current Optimal Power Flow Problem Alvaro Lorca, Georgia Institute of Technology, alvarolorca@gatech.edu, Andy Sun We present an adaptive robust optimization model for the alternating current optimal power flow problem (ACOPF) under uncertainty in renewable power availability and the active and reactive power injections at demand nodes. We will discuss solution methods and the performance of the approach proposed through computational experiments. 2 - Comparison Of Stochastic Programming And Robust Optimization For Risk Management In Energy Generation Ricardo M Lima, KAUST, Thuwal, Saudi Arabia, ricardo.lima@kaust.edu.sa, Sabique Langodan, Ibrahim Hoteit, Omar Knio, Antonio J. Conejo In this talk, we address the optimal self-scheduling and market involvement of a virtual power plant (VPP) by using three different methods. The VPP faces a decision-making problem with uncertainty in the wind power and electricity prices forecast. We define this problem using a risk-averse stochastic programming model, a robust optimization model, and with a new hybrid robust- stochastic formulation. We analyze these methods from the point of view of formulations, uncertainty quantification and risk, decomposition algorithms, and computational performance. Furthermore, we compare the impact of the risk measures and their parameterizations on the results obtained with the three methods. 3 - Risk-constrained Optimal Power Flow With Moment And Unimodality Information Bowen Li, University of Michigan, Ann Arbor, MI, United States, libowen@umich.edu, Ruiwei Jiang, Johanna Mathieu We propose a risk-constrained optimal power flow (OPF) problem with uncertain renewables. Using historical data and domain knowledge, we incorporate moments of the renewable forecast errors and assume unimodality to derive reformulations and approximations based on semidefinite programs and second- order cone programs, and evaluate them on IEEE test systems. 4 - An Efficient Robust Solution To The Two-stage Stochastic Unit Commitment Problem Ignacio Blanco, PhD Student, Technical University of Denmark, Kgs. Lyngby, 2800, Denmark, igbl@dtu.dk, Juan M Morales Gonzalez This paper proposes a reformulation of the scenario-based two-stage unit commitment problem under uncertainty that allows finding unit-commitment plans that perform reasonably well both in expectation and for the worst case realization of the uncertainties. The proposed reformulation is based on partitioning the sample space of the uncertain factors by clustering the scenarios that approximate their probability distributions. It is, furthermore, very amenable to decomposition and parallelization using a column-and-constraint generation procedure. WD05 101E-MCC Wildland Fire Management I - Suppression Sponsored: Energy, Natural Res & the Environment II Forestry Sponsored Session Chair: Eghbal Rashidi, Mississippi State University, Industrial & Systems Engineering, Mississippi State University, MS, 39762, United States, er442@msstate.edu 1 - Vulnerability Analysis Of The Initial Attack In Suppressing The Worst Case Wildfires Eghbal Rashidi, Mississippi State University, er442@msstate.edu, Hugh Medal In this research, we perform a quantitative gap analysis between available capacity for responding to wildfires and the estimated capacity needs for responding to a worst case scenario wildfire. We model the problem as a Stackelberg game using a bilevel max-min MIP model. We use the model to evaluate the impact of the worst-case wildfire, i.e., the arrangement of ignition

457