报告内容： In this paper, we study a cyclic inventory routing problem (CIRP). The traditional exact methods for the inventory routing problem (IRP) use an arc-based formulation (also known as two-index flow formulation), in which a variable represents a possible vehicle flow between a pair of customers. In this research, we propose a schedule-based model (SBM), in which a variable represents a possible one-day schedule for any vehicle. This model can be considered as a Dantzig-Wolfe decomposition of the arc-based model. We also propose a set of new valid inequalities to tighten the linear relaxation bound of SBM. To solve SBM efficiently, we develop a column generation algorithm, in which only attractive vehicle schedules are generated. Our computational results on five real-life test cases show that the LP relaxation of SBM is tight. SBM can obtain near optimal solutions to very large real-life test cases within a reasonable time, and average IP-LP gaps of SBM are within 5% for ML policy and 7% for OU policy respectively.
报告人简介： Zhe Liang, Professor in Department of Management Science & Engineering, TongjiUniversity, ShanghaiChina. He received his Bachelor and Master in Department of Computer Engineering and Department of Industrial & Systems Engineering from National University of Singapore in 2001 and 2003 respectively, and Ph.D. in Department of Industrial & Systems Engineering from Rutgers University in 2011. His research interests are related to the design and implementation of exact and heuristic algorithms for large-scale combinatorial optimization problems in supply chain management, transportation, telecommunication, and computational biology. He published over 20 referred research articles, including his papers appeared in journals such as Transportation Science, Transportation Research Part B, INFORMS Journal on Computing, and others. He is the recipient of The China National Science Fund for Excellent Young Scholars in 2014. |
