Rollout Policies for Dynamic Solutions to the Multivehicle Routing Problem with Stochastic Demand and Duration Limits

被引:85
作者
Goodson, Justin C. [1 ]
Ohlmann, Jeffrey W. [2 ]
Thomas, Barrett W. [2 ]
机构
[1] St Louis Univ, John Cook Sch Business, Dept Operat & Informat Technol Management, St Louis, MO 63108 USA
[2] Univ Iowa, Tippie Coll Business, Dept Management Sci, Iowa City, IA 52242 USA
关键词
VEHICLE; ALGORITHMS;
D O I
10.1287/opre.1120.1127
中图分类号
C93 [管理学];
学科分类号
12 ; 1201 ; 1202 ; 120202 ;
摘要
We develop a family of rollout policies based on fixed routes to obtain dynamic solutions to the vehicle routing problem with stochastic demand and duration limits (VRPSDL). In addition to a traditional one-step rollout policy, we leverage the notions of the pre- and post-decision state to distinguish two additional rollout variants. We tailor our rollout policies by developing a dynamic decomposition scheme that achieves high quality solutions to large problem instances with reasonable computational effort. Computational experiments demonstrate that our rollout policies improve upon the performance of a rolling horizon procedure and commonly employed fixed-route policies, with improvement over the latter being more substantial.
引用
收藏
页码:138 / 154
页数:17
相关论文
共 37 条
[1]   A paired-vehicle recourse strategy for the vehicle-routing problem with stochastic demands [J].
Ak, Aykagan ;
Erera, Alan L. .
TRANSPORTATION SCIENCE, 2007, 41 (02) :222-237
[2]  
American Trucking Association, 2009, ALL ALL POL CLIM LEG
[3]  
[Anonymous], 2001, TION ENGRG
[4]  
[Anonymous], 2007, Approximate Dynamic Programming: Solving the Curses of Dimensionality (Wiley Series in Probability and Statistics)
[5]   Rollout Algorithms for Combinatorial Optimization [J].
Bertsekas D.P. ;
Tsitsiklis J.N. ;
Wu C. .
Journal of Heuristics, 1997, 3 (3) :245-262
[6]  
Bertsekas D. P., 2000, DYNAMIC PROGRAMMING, VI
[7]   Revenue management in a dynamic network environment [J].
Bertsimas, D ;
Popescu, I .
TRANSPORTATION SCIENCE, 2003, 37 (03) :257-277
[8]   Computational approaches to stochastic vehicle routing problems [J].
Bertsimas, D ;
Chervi, P ;
Peterson, M .
TRANSPORTATION SCIENCE, 1995, 29 (04) :342-352
[9]   An approximate dynamic programming approach to multidimensional knapsack problems [J].
Bertsimas, D ;
Demir, R .
MANAGEMENT SCIENCE, 2002, 48 (04) :550-565
[10]   A new generation of vehicle routing research: Robust algorithms, addressing uncertainty [J].
Bertsimas, DJ ;
SimchiLevi, D .
OPERATIONS RESEARCH, 1996, 44 (02) :286-304