2016 INFORMS Annual Meeting Program

TC68

INFORMS Nashville – 2016

3 - Classification Based Approach To Nanorod Segmentation In Scanning Electron Micrographs Mostafa Gilanifar, Florida State University, Tallahassee, FL, United States, mg14m@my.fsu.edu, Abhishek Shrivastava Estimation of nanorod morphology from scanning electron micrographs requires extracting nanorods from the image. This foreground-background segmentation of nanorods is challenging due to several reasons - low signal to noise ratio, high degrees of nanorod overlap and shape similarity. In this talk, we present a classification-based approach to the segmenation problem. We use a decomposition-based approach to identify nanorod and background image patterns, which can be discriminated accurately using a trained classifier. We demonstrate the accuracy of the approach through several examples. 4 - An Order-invariant Cholesky-log-garch Model For Multivariate Financial Time Series Xinwei Deng, Associate professor, Virginia Tech, 211 Hutcheson Hall, 250 Drillfield Drive, Blacksburg, VA, 24060, United States, xdeng@vt.edu, Xiaoning Kang, Kam-Wah Tsui, Mohsen Pourahmadi Accurate estimation of time-varying covariance matrices is of great importance in the analysis of financial data. Most existing models are known to break down at the estimation stage for dimensions larger than ten or so. In this work, we propose a novel order-invariant Cholesky-log-GARCH model for estimating the covariance matrix of multivariate time series based on a random sample from a population of all possible permutations of the p variables. The ensuing methodology not only provides accurate estimation, but also gives accurate prediction at future time points. The merits of the proposed method are illustrated through three real financial data sets in comparison with conventional methods. TC68 Mockingbird 4- Omni Reliability Evaluation and Optimization from Complex Systems II Sponsored: Quality, Statistics and Reliability Sponsored Session Chair: Eunshin Byon, University of Michigan, College Station, MI, United States, ebyon@umich.edu Co-Chair: Qingyu Yang, Wayne State University, Detroit, MI, United States, qyang@wayne.edu 1 - Reliability Estimation Of Systems With Spherically Distributed Units Jingbo Guo, Rutgers, The State University of New Jersey, guojingbochina@gmail.com, Elsayed A. Elsayed Spherically distributed systems are emerging in a diverse range of industries such as aerospace, nuclear, military and oceanography. We refer this kind of units on a spherical arrangement as k-n-i: G balanced systems. In such a system, n pairs of units are distributed evenly on i vertical planes in a sphere. The system is considered functioning when at least k out of n pairs of units operate properly while satisfying the system’s balance requirement. In this presentation, we introduce an efficient and general algorithm to estimate the reliability of such systems for any range of k, n, i. We also present a numerical example to illustrate the algorithm. 2 - Condition-based Selective Maintenance Optimization For A Large-scale System Young Myoung Ko, Pohang University of Science and Technology, youngko@postech.ac.kr, Eunshin Byon We extend our previous study on condition-based maintenance optimization that schedules maintenance activities in a large-scale system comprising identical units. Our previous method was based on the assumption that each maintenance activity renews all units, which made the analysis tractable by taking advantage of the results from the renewal theory, but indeed restricts the applicability of the method. In this talk, we present a new approach that relaxes the renewal assumption and finds the optimal thresholds that trigger maintenance operations to repair a subset of units (i.e., highly deteriorated or failed units) in a system. 3 - A Two-stage Structural Degradation Modeling In Sparse Datasets Abdallah Chehade, University of Wisconsin - Madison, chehade@wisc.edu, Kaibo Liu Degradation modeling has become essential to the field of condition monitoring for better logistics and decision-making. Existing approaches assume there exists a high-quality data-rich environment. However, in many scenarios, the provided dataset is sparse with limited observations. For example, patients tend to skip their semi-annual clinic visits and result in a highly sparse dataset. To fill the literature gap, we propose a novel two-stage approach for structural degradation modeling in sparse datasets. Both simulation and case studies that involve a dataset (ADNI) for Alzheimer disease were used to numerically evaluate and compare the performance of the proposed methodology.

4 - Title: Resistance Level Determination At The Reliability-based System Design Qiyun Pan, University of Michigan, qiyun@umich.edu, Eunshin Byon In order to provide a guideline for choosing appropriate design parameters to meet a required level of system reliability, resistance level determination becomes crucial in many applications. At the design stage, resistance level can be estimated using stochastic simulations, and the resistance level estimation can be formulated as a statistical quantile estimation problem. We present a new adaptive importance sampling algorithm to improve the estimation accuracy, given a computational budget. TC69 Old Hickory- Omni Airports, Runways, and Descents Sponsored: Aviation Applications Sponsored Session Chair: Emad Alharbi, NJIT, 8 Gordon Cir, Parsippany, NJ, 07054, United States, eaa3@njit.edu 1 - An Airport Scheduling Mechanism Based On Efficiency, Equity And On-time Performance Objectives Alexandre Jacquillat, Assistant Professor of Operations Research, Carnegie Mellon University, 5000 Forbes Ave, Pittsburgh, PA, 15213, United States, ajacquil@andrew.cmu.edu, Vikrant Vaze Airport congestion can be mitigated through scheduling interventions that control imbalances between peak-hour demand and capacity. We design and optimize a non-monetary scheduling mechanism that starts with scheduling inputs from the airlines and airport capacity estimates, and that reschedules flights based on efficiency, inter-airline equity and on-time performance objectives. Theoretical and computational results suggest that large equity gains can be achieved at no, or small, losses in efficiency, and that accounting for airline preferences can enhance the mechanism outcomes. 2 - Integration Of Airport Runway And Taxi Planning Giuseppe Sirigu, ETS Ingeniería Aeronáutica y del Espacio, Dep Matemática Aplicada Ingenieria Aeroespaci, Plaza Cardenal Cisneros 3, Madrid, 28040, Spain, angel.marin@upm.es, John-Paul Clarke, Angel Marin We develop and compare network flow-based and Monte Carlo sampling-based deterministic algorithms for the joint optimization of runway and taxiway operations considering required separation minima and environmental concerns. The objective is to minimize the movement time and/or taxi delay considering the desired takeoff time windows and earliest pushback times. The algorithms are compared using numerical simulations that replicate real-word airport operations. 3 - Departure Queue Management - A Data Driven Analysis Marc Rose, Senior Operations Research Analyst, MCR Federal, LLC, 600 Maryland Ave, 306E, Washington, DC, 20024, United States, mrose@mcri.com At the FAA the Terminal Flight Data Manager (TFDM) is composed of many capabilities. A major component is the Departure Queue Manager (DQM) which is designed to shift aircraft taxi-out delay away from the runway queue to either the gate (preferred) or a designated waiting area, thus saving fuel. In this paper I discuss the databases and calculations applied to estimate the amount of time that can be shifted from the queue. This will include some discussion of the programming and constraints required to capture some of the uncertainties in the concept 4 - Continuous Descent Arrival Adoption During High Traffic Periods: A Data-driven And Predictive Modeling Approach Emad Alharbi, PhD Candidate, New Jersey Institute of Technology, Newark, NJ, 07102, United States, eaa3@njit.edu, Layek Abdel-Malek This study investigates Continuous Descent Arrival (CDA) adoption during high traffic levels periods. We utilize data-driven system approach and predictive analytics to build an online CDA predictive model for an enhanced Air Traffic Management (ATM) procedures as well as an efficient CDA adoption. A Hierarchical Clustering Analysis (HCA) is performed to aggregate data from offline flight tracking logs and Meteorological Aviation Reports (METARs) at selected U.S. airports. The analysis facilitates the visualization of descent profiles and assists in developing a predictive model for CDA instances using Decision Trees with AdaBoost and Support Vector Machines (SVM).

326