美国数学建模竞赛特等奖论文——2004 B O KalmanQueue An Adaptive Approach to Virtual Queueing【数学建模】.pdfVIP

KalmanQueue 337 KalmanQueue: An Adaptive Approach to Virtual Queueing Tracy Clark Lovejoy Aleksandr Yakovlevitch Aravkin Casey Schneider-Mizell University of Washington Seattle, WA Advisor: James Allen Morrow Summary QuickPass (QP) is a virtual queueing system to allow some theme-park guests to cut their waiting time by scheduling a ride in advance. We propose innovative QP systems that maximize guest enjoyment. Only a small portion of guests can effectively use QP, and a good system maximizes this group subject to the constraints that regular users are not significantly affected and maximum waiting time for QP users is small. We define and test a simple model for single-line formation and then de- velop two QP systems, GhostQueue and KalmanQueue. GhostQueue is intu- itive and simple but would be far from optimal in practice. We then propose that the best model is one that adapts to its environment rather than trying to enforce rigid parameters. We implement KalmanQueue, a highly adaptive sys- tem that uses an algorithm inspired by the Kalman filter to adjust the number of QPs given today based on the maximum length of the QP line yesterday, while filtering out random noise. We simulate the KalmanQueue system with a C++ program and randomized input from our line-formation model. This system quickly converges to a nearly optimal solution. It is, however, sensitive to some parameters. We discuss the expected effectiveness of the system in a real environment and conclude that KalmanQueue is a good solution. The UMAPJoutrnal 25 (3)(2004) 337-352. ®Copyright2004 by COMAP, Inc. All rights reserved. Permission to make digital or hard copies of part or a