Informs Annual Meeting Phoenix 2018

INFORMS Phoenix – 2018

WD34

4 - Optimization Problem of Parcel Delivery Coordinating a Truck with Drones Xin Wang, Tsinghua University, Shenzhen, 518055, China, Mingyao Qi With the emergence of technologies, a new distribution method of deploying the unmanned aerial vehicles or drones to support the parcel delivery gradually exerts its advantages. A new idea of using a drone in conjunction with a traditional delivery truck to distribute parcels is discussed in earlier literatures. Different from the existing work, we propose a new distribution mode of a truck with several drones and provide a mathematical programming model aiming to find the optimal routing and scheduling of the multiple vehicles. 5 - An Adaptive Large Neighborhood Search Heuristic for the Time Dependent Profitable Pickup and Delivery Problems with Time Windows Peng Sun, Kuhne Logistics University, Grosser Grasbrook 17, Hamburg, 20457, Germany, Lucas P. Veelenturf, Mike Hewitt, Tom van Woensel In this paper, we study the time-dependent profitable pickup and delivery problem with time windows. In this problem, each request consists of a pickup and delivery location, a profit is collected from the visit to its pickup. A limited amount of vehicles with a capacity limit are available. The profit of a request can be collected at most once. Time-dependent travel times are considered to capture road congestion. The objective is to determine a set of tours that maximize the difference between the collected profits and the total traveling cost. An adaptive large neighborhood search is proposed. Results show its effectiveness in finding good-quality solutions on the instances with up to 75 requests. Game Theory III Contributed Session Chair: Timothy Murray, University of Illinois, Champaign, IL, 61820, United States 1 - Schrodinger Nim Mark J. van den Bergh, Leiden University, Ravenhorst 59, Leiden, 2317AG, Netherlands In the field of combinatorial game theory, many beautiful results are known for two-player, deterministic, perfect information games, of which Nim is a prime example. In this talk, we will discuss a non-perfect information variant of Nim. Employing results from non-cooperative game theory, we assess the probability of the starting player being the winner for various sizes of piles, and compare the results to the classic version of Nim. 2 - What if Hotelling Firms Can Mass Customize Alireza Yazdani, University of Oregon, 1208 University Street, Eugene, OR, 97403, United States, Eren Basar Cil, Michael Pangburn We study product-design and price competition between two mass-customizing firms that serve consumers with varying tastes and finite reservation prices on a Hotelling linear city. By comparing equilibrium results in settings with and without mass customization, we show that mass customizers prefer markets exhibiting extreme (high or low) levels of consumers’ fit sensitivity, whereas traditional firms are better off when facing moderate fit sensitivity. We also establish that competition with mass customization may lead to lower profits and consumer surplus, suggesting that regulators should closely evaluate whether to facilitate industry investments in customization technologies. 3 - Effects of Limited Trust on the Price of Anarchy of Random Repeated Games Timothy Murray, University Illinois at Urbana-Champaign, Champaign, IL, 61820, United States, Jugal Garg, Rakesh Nagi We examine Leader-Follower Repeated Games of 2 players on payoff matrices randomly generated from arbitrary. We explore three concepts of Limited Trust (rather than altruism) and their effects on the Price of Anarchy, both theoretical and empirical. n WD34 North Bldg 223

n WD35 North Bldg 224A Data-driven Analysis for Air Transport Sponsored: Aviation Applications Sponsored Session Chair: Yulin Liu, University of California-Berkeley, 107 McLaughlin Hall, Berkeley, CA, 94709, United States 1 - Predicting Actual Aircraft Trajectory with Deep Mixture Density Recurrent Neural Networks Yulin Liu, University of California-Berkeley, 107 McLaughlin Hall, Berkeley, CA, 94709, United States, Mark M. Hansen, Michael O. Ball, David J. Lovell Reliable aircraft trajectory prediction, whether in a real-time setting or for analysis of counterfactuals, is important to the efficiency of the aviation community. We propose an end-to-end deep learning framework that consists of a Long Short-Term Memory (LSTM) encoder and a mixture density LSTM decoder to predict aircraft trajectories based on last filed flight plans, wind conditions, and convective weather. The encoder network extracts features from the flight plans, and the decoder network learns the joint distribution of the state of an aircraft from meteorological conditions. Beam search and filtering algorithms are used to stabilize the prediction results in the inference phase. 2 - Physics Based Learning for Simulation and Prognostic of Aircraft Dynamical System Yang Yu, Postdoctoral Research Associate, Arizona State University, Tempe, AZ, United States, Houpu Yao, Yongming Liu This study proposes the concept of physics-based learning, a hybrid approach based on data-driven learning and physical models, as a computationally efficient method for the simulation of aircraft dynamics. The physics-based learning integrates the underlying physics of dynamical systems into learning models such as neural networks to reduce the training and simulation costs. The application of physics-based learning for simulating aircraft dynamics is demonstrated using a recently introduced physics-aware network known as the deep residual recurrent neural network (DR-RNN) on a Boeing 747-100 aircraft. 3 - Denoising and Risk Identification of Bird Strikes in Civil Aviation Based on Complementary Ensemble Empirical Mode Decomposition Chen Zhang, CASTC, Beijing, China Bird strikes are considered as one of the important factors threatening to the safety of civil aviation. In this paper, a new application based on complementary ensemble empirical mode decomposition (CEEMD) was introduced. With the data of bird strikes from 2004 to 2013, it was split into a set of intrinsic mode functions (IMFs) and a residue. During the process of CEEMD decomposition, more features had been extracted that bird strikes were prone to influence by intermittent cyclicity. What’s notable is that the MHIs always appeared in September. These findings provided a theoretical basis to early warning and better measures taken for bird strikes risk management. n WD36 North Bldg 224B Practice- Inventory Management Contributed Session Chair: Yixuan Xiao, City University of Hong Kong, 83 Tat Chee Avenue, Kowloon, 1, Hong Kong 1 - Capacity Expansion with a Bundled Supply of Capacity Attributes Mohammad Ebrahim Arbabian, University of Washington, Seattle, WA, 98105, United States, Shi Chen, Kamran Moinzadeh We study the well-known problem of expanding capacity of server attributes in a cloud company where supply of attributes is bundles. We consider a cost minimization problem in a continuous review, finite horizon setting. Furthermore, the best server configurations to be deployed each cycle are studied 2 - Managing Perishable Inventory Systems with Multiple Age-differentiated Demand Classes Shouchang Chen, Zhejiang University, Hangzhou, China, Yanzhi Li, Yi Yang, Weihua Zhou In this paper, we consider a periodic-review inventory system with perishable products. In the market, there are different demand classes; each can be characterized by its different lost-sale cost and freshnessrequirement. By establishing some new properties of multimodularity, we partially characterize the structure of the optimal policy. Based on optimality analysis, we design several efficient approximation approaches.Numerical studies show that our heuristic policy outperforms the traditional heuristics proposed by prior literature.

506