2016 INFORMS Annual Meeting Program

369 / 561

INFORMS Nashville – 2016
369
WA18
106A-MCC
DMA Data-Driven Models
Contributed Session
Chair: Nuo Xu, University of Alabama at Birmingham, 5720 11th
Avenue South, Birmingham, AL, 35222, United States,
1 - Remaining Useful Life Prediction Of Lithium Ion Batteries Using A
Novel Degradation Model
Fangfang Yang, City University of Hong Kong, Hong Kong, China,
Kwok-Leung Tsui
Some lithium-ion battery materials show two-phase degradation behavior, such
as lithium nick manganese cobalt oxide (NMC) cells. To predict remaining useful
life (RUL) for these types of batteries, a model-based Bayesian approach is
propose. First, a novel degradation model is developed to capture the degradation
trend of NMC batteries. Next, a particle filtering-based prognostic method is
incorporated into the model to estimate possible degradation trajectories of the
batteries. The effectiveness of the developed method is verified using our
experimental data. The results indicate that the proposed prognostic method can
achieve high prediction accuracies at an early stage of life.
2 - Process Monitoring And Diagnosis Of Hot Rolled Trip Based On
Regression Coefficients Of Batches
Fei He, University of Science and Technology Beijing, China,
Beijing, China,
First of all, regression model between process parameters and product quality data
is established. And then regression coefficients are used for process monitoring
and diagnostics. In this paper the model based on partial least squares is bulit
between process variables and width of finishing hot rolling, and regression
coefficients of all batches are obtained that is used for process monitoring and
diagnostics. Experiments on simulated data sets and real data sets show that can
effectively locate the important abnormal process parameters.
3 - Data Science And The Liberal Arts Curriculum
Anna Engelsone, Maryville College, Maryville, TN, United States,
This paper draws on over ten years of experience practicing DMA in an industry
setting and teaching data science concepts to students ranging from 8th graders to
MBAs. Our main interest is in incorporating DMA into the liberal arts curriculum.
Liberal arts colleges are uniquely positioned to produce versatile data
professionals with the ability to ask the right questions, consider the social
implications of their work, and communicate their findings effectively to different
audiences. We discuss the challenges of introducing DMA to undergraduates and
present examples of in-class exercises, homework problems and research projects
suitable for students of different levels and backgrounds.
4 - A Measure Of General Functional Dependence Among Multiple
Continuous Variables
Nuo Xu, University of Alabama at Birmingham, 5720 11th Avenue
South, Birmingham, AL, 35222, United States,
Xuan Huang
Existing measures in the literature that are specifically concerned with testing and
measuring independence between two continuous variables are all based on
examining the definition of independence. In a previous paper of ours, we
construct a new measure that uses the absolute value of first difference on
adjacent ranks of one variable with respect to the other. This measure captures
the general functional dependence between two variables. Here, we are
presenting the method of generalizing this measure to capture functional
dependence among N variables and some preliminary results of its application in
variable interaction detection and variable selection.
WA19
106B-MCC
Uncertainty in Engineered Networks
Sponsored: Computing
Sponsored Session
Chair: Russell Bent, Los Alamos National Laboratory, Los Alamos
National Laboratory, Los Alamos, NM, 00000, United States,
1 - Optimal Robust Battery Operation
Shuoguang Yang, Columbia University,
We present formulations, algorithms and computational results on mult-time
period problems involving battery operation. In this context, batteries are used to
compensate for errors in forecasts for renewable power generation. We model
uncertainty sets using the uncertainty budgets model, and we describe efficient
implementations. Joint work with D. Bienstock, G. Munoz and C. Matke.
2 - Unit Commitment With N-1 Security And Wind Uncertainty
Kaarthik Sundar, Texas A&M,
Harsha Nagarajan, Miles Lubin, Sidhant Misra, Russell Bent,
Line Roald, Daniel Bienstock
As wind energy penetration rates continue to increase, a major challenge facing
grid operators is the question of how to control transmission grids in a reliable
and a cost-efficient manner. The stochasticity of wind forces an alteration of
traditional methods for solving the day-ahead unit commitment problem. To
address these questions, we present an N-1 Security and Chance-Constrained
Unit Commitment that includes the modeling of generation reserves to respond to
wind fluctuations and tertiary reserves to account for single component outages.
We develop a benders decomposition algorithm to solve the problem to optimality
and present a detailed case study on the IEEE RTS-96 three-area system.
3 - Efficient Dynamic Compressor Optimization In Natural Gas
Transmission Systems
Pascal Van Hentenryck, University of Michigan,
The growing reliance of electric power systems on gas-fired generation to balance
intermittent sources of renewable energy has increased the variation and volume
of flows through natural gas transmission pipelines. Adapting pipeline operations
to maintain efficiency and security under these new conditions requires
optimization methods that account for transients and that can quickly compute
solutions in reaction to generator re-dispatch. This talk presents an efficient
scheme to minimize compression costs under dynamic conditions where
deliveries to customers are described by time-dependent mass flow.
WA20
106C-MCC
Mining Qualitative Attributes to Assess
Corporate Performance
Invited: Tutorial
Invited Session
Chair: Ananda Swarup Das, IBM India Research Labs, India,
New Delhi, 1, India,
1 - Mining Qualitative Attributes To Assess Corporate Performance
Aparna Gupta, Rensselaer Polytechnic Institute,
110 Eighth Street, Troy, NY, 12180, United States,
Ananda Swarup Das, L Venkata Subramaniam, Gagandeep Singh
We present an overview of systems and methods to track ongoing events from
sources such as corporate filings, financial articles, expert or analyst reports, press
releases, customers’ feedback and news articles that have an effect on corporate
performance. In this paper we discuss text analytics and sentiment mining
approaches to determine quantitative attributes that can be an indicator of
corporate performance. For example, strengths, weaknesses, opportunities and
threats (SWOT) analysis is a well-known structured planning method widely
applied to identify the factors determining success or failure of an enterprise. This
analysis can be strongly indicative of the business or financial health of the
enterprise. It can provide broader indicators for the firm’s business environment,
in terms of ease of doing business in the country, government policies helping (or
hurting) business environment.
WA21
107A-MCC
Chronic Disease Management
Sponsored: Health Applications
Sponsored Session
Chair: Vedat Verter, McGill University, 1001 rue Sherbrooke Ouest,
Bronfman Building, Montreal, QC, H3A 1G5, Canada,
Co-Chair: Michael Klein, McGill University, McGill, Montreal, QC,
Canada,
1 - Chronic Disease Management And The Role Of Incentives
Christian Wernz, Virginia Tech, , Hui Zhang
Chronic diseases can be prevented by changing the behavior of patients and
physicians. Incentives are one of the mechanisms to motivate such change. We
present a two-player, multi-period model in which patients and physicians jointly
decide on prevention activities. The physician-patient interaction is modeled as a
general-sum stochastic game with switching control structure. The Health Belief
Model (HBM) is incorporated to capture behavioral aspects. We illustrate our
modeling approach by applying it to a coronary heart disease cases study. Result
show how and to what extent a re-alignment of incentives can improve chronic
disease management initiatives.
WA21