2016 INFORMS Annual Meeting Program

WE64

INFORMS Nashville – 2016

WE62 Cumberland 4- Omni Passengers and Parts: Analytics and Machine Learning in Aviation Sponsored: Aviation Applications Sponsored Session Chair: Catherine Cleophas, RWTH Aachen University, Kackertstr. 7, Aachen, 52072, Germany, catherine.cleophas@rwth-aachen.de Co-Chair: Phillip Mah, Boeing, Commerce, Richmond, BC, V6V 2L1, Canada, phillip.mah@aeroinfo.com 1 - Predictive Maintenance from Analysis Of Airplane Sensor Data Phillip Mah, Boning, Commerce, Richmond, BC, V6V 2L1, Canada, phillip.mah@aeroinfo.com, Ruiwei Jiang, Dawen Nozdryn-Plotnicki Unscheduled maintenance drives 10% of the annual operational cost to airlines worldwide. Predictive Maintenance could reduce those costs, particularly when synchronized with airline’s operations. By using engineering expertise, statistics and machine learning on aircraft sensor and fault data, as well as analysis of an airline’s flight and maintenance schedule, we detect impending issues on the aircraft and suggest maintenance tasks in accordance with the prediction and an airline’s working rhythm. These predictive maintenance tasks will increase reliability and reduce unscheduled maintenance. 2 - Applications Of Text Mining Techniques To Fleet Health And Maintenance Data Textual data is one of the richest data sources for fleet health and maintenance analytics. Taking advantage of these information is the key for optimizing airline’s and Boeing’s business. Due to its large volume and highly unstructured nature, however, its full potential is rarely leveraged. Advanced Analytics group works on text analytics projects with airlines in the areas such as reliability, maintenance program and Boeing customer support. Case studies on how we help businesses by applying natural language processing, machine learning and visualization techniques will be presented. 3 - On The Flight Choice Behaviour Of Business Purpose Passengers In The Australian Domestic Market Cheng-Lung Wu, Associate Professor, UNSW Australia, School of Aviation, UNSW Australia, Kensington, NSW 2052, Australia, c.l.wu@unsw.edu.au, Hanson So This paper examined the flight choice behaviour difference of business-purpose passengers who work in small and medium enterprises (SMEs,) and those in non- SMEs. Statistics show that SME business passengers tend to fly less, are more price-sensitive, and derive less satisfaction in flying with full-service carriers if they have previously flown with low-cost carriers. Discrete choice models show that fewer flight service attributes are significant on shorter flights. Flight comfort attributes have a larger significance on inbound flights. Attribute non-attendance (ANA) is above 55% for all tested attributes except fare; not all attributes are perceived equally by business passengers. Bingjing Yu, Boeing Vancouver, 1146 Homer Street, Vancouver, BC, V6B 2X6, Canada, bingjing.yu@aeroinfo.com, Gaku Tobinobu, Candice Chan, Ehsan Nobakht

2 - A Robust Optimization Approach For Ambulance Location Problem Hozumi Morohosi, National Graduate Institute for Policy Studies, morohosi@grips.ac.jp, Takehiro Furuta This work studies an application of robust optimization to a cooperative covering model with focusing on ambulance location problem. We propose a procedure for defining the uncertain set of parameters in the problem based on actual data in a Bayesian way. Then we bring out a robust version of cooperative covering problem and give a solution analysis with some numerical examples. 3 - Generalized Weighted Benefit And Maximal Expected Covering Location Problem Daisuke Watanabe, Tokyo University of Marine Science and Technology, daisuke@kaiyodai.ac.jp, Richard Church The purpose of this study is to analyze the optimal location model for Counter- Piracy Surveillance System in Somalia using Weighted Benefit and Expected Coverage Model based on Maximal Covering Location Problem. WE64 Cumberland 6- Omni Vector Optimization: Algorithms and Applications Sponsored: Multiple Criteria Decision Making Sponsored Session Chair: Firdevs Ulus, Bilkent University, Ankara, Turkey, firdevs@bilkent.edu.tr 1 - Multiobjective Risk-averse Two-stage Stochastic Programming Cagin Ararat, Assistant Professor, Bilkent University, Ankara, 06800, Turkey, cararat@bilkent.edu.tr, Ozlem Cavus Risk-averse two-stage stochastic programming is concerned with the minimization of a risk measure of a random cost function over the feasible choices of a deterministic and a random decision variable. We study the multiobjective version of this problem in which case the cost function is vector- valued and its risk is quantified via a set-valued risk measure. Although the resulting problem has a set-valued objective function, we reformulate it as a convex vector optimization problem and propose a customized version of Benson’s algorithm to solve it. In particular, we develop duality-based cutting- plane type methods to solve the scalar subproblems appearing in Benson’s algorithm. 3 - Convex Vector Optimization Problems: Unboundedness Firdevs Ulus, Bilkent University, firdevs@bilkent.edu.tr In order to solve convex vector optimization problems (CVOPs), Benson type algorithms have been proposed recently. These algorithms work well if the feasible region is compact which implies that the problem is bounded and a ‘finite weak epsilon-solution’ to the problem exists. However, in many applications, the feasible region of the problem does not necessarily compact and the problem may be unbounded. In this talk, the aim is to discuss the following: 1- Is there a finite solution concept for unbounded convex optimization problems? 2- If this is the case, are there conditions which guarantee the existence of such a solution? 3- Corinna Krüger, Georg August University Göttingen, Göttingen, Germany, ckrueger@math.uni-goettingen.de, Anita Schöbel, Margaret M Wiecek We investigate linear multiobjective optimization problems with uncertain right hand sides of the constraints which are modeled as the elements of a polyhedral uncertainty set. Generalizing Ben-Tal and Nemirowski’s definition of the robustness gap (RG) for the single objective case to the multiobjective case, we quantify the gap in two ways. We develop a quadratic program whose objective value corresponds to the RG. We also relate the RG to the distance beween appropriate Pareto sets. Can we extend the existing algorithms to unbounded CVOPs? 2 - Quantification Of The Robustness Gap For Uncertain Multiobjective Optimization Problems

WE63 Cumberland 5- Omni Urban Operations Research Sponsored: Location Analysis Sponsored Session

Chair: Daisuke Watanabe, Tokyo University of Marine Science and Technology, 2-1-6 Etchujima, Koto-ku, Tokyo, 135-8533, Japan, daisuke@kaiyodai.ac.jp 1 - Coverage Modeling In Public Street Lighting Alan T. Murray, University of California at Santa Barbara, Santa Barbara, CA, 93106, United States, amurray@ucsb.edu, Xin Feng The spatial distribution of public area lighting in an urban region greatly influences human activities and safety, yet is costly to provide. This paper details a coverage optimization model for nighttime light provision, enabling benefits and impacts to be taken into account. Application results for an urban area are presented and discussed.

503