Abstract: “Reasonable logistics distribution”is the essential guarantee with agile logistics. But the vehicle routingproblem, as the traditional model for logistics distribution, fails behind theneed of agile logistics. By analysis of the characters of agile logistics anddifferent kinds of information, a new model for agile vehicle routing problemunder complex information, which includes both deterministic information andstochastic one, is given. Utilizingknowledge of queue theory, combinatorial optimization and geometry probability,the dynamic policyis given as the optimal solution. The probability andframework of the new way is given. Numerical experiments are conducted toinvestigate the new model, and to assess the effectiveness of the dynamicpolicy by comparing with the traditional way.
Keywords:Agile logistics, Complex information,Dynamic policy, Stochastic information, Two-attempt approach, Vehicle routingproblem
1Introduction
The preliminaryproblem for agile logistics [1]is how to satisfy customers in thetime window. “Reasonable logistics distribution” is considered as one of the essential guarantee with agile logistics.
The model for “logistics distribution”, named theVehicle Routing Problem (VRP), was first built by Dantzig and Ramer in 1959[2].In the classical VRP, all kinds of information are assumed to be determinate,in practice, planner of routes must meet with stochastic information accordingwith statistical law frequently. Theapplicability of methods for the VRP is hampered by the uncertainty.
VRP under stochastic information [3]isproposed by Hochberg and Tamhane (1987), Bechnofer and Santner (1995), and Hsu(1996) for cases where the set of feasible solutions is small. According to thetraditional ways, approximate deterministic formulations are employed whenuncertain values are replaced by expected values or by percentiles, but itcannot always be done accurately and realistically.