Research On Multi-depot Vehicle Routing Opimization Problem

With rapid development of the economics globalization, the company orenterprise scale unceasingly expands, in order to make limited resources optimalallocation, improve the quality and efficiency of economic operation. Multi-depotvehicle routing problem (MDVRP) has potential applications in many fields. Inreal-life world, there have many uncertain factors. Therefore, multi-depot vehiclerouting problem with time window base on credibility theory is discussed in this paper.This paper is summarized as follows:1) In order to deal with some uncertain parameters in real-life. For multi-depotvehicle routing problem with time window, we assume that the travel time isfuzzy variable based on credibility theory. Three types of model areestablished, which are multi-objective programming expected value model,chance constrained multi-objective programming model and chanceconstrained goal programming model.2) The multi-depot vehicle routing problem is turned as single-depot vehiclerouting problem. For chance constrained multi-objective programming model,A new algorithm which integrates partheno genetic algorithm and fuzzysimulation technology is presented.3) The bi-level programming expected value model and chance constrainedbi-level programming model for multi-depot vehicle routing problem areestablished based on credibility theory.4) The chance constrained bi-level programming model is solved by a bi-levelhybrid intelligence algorithm which combines partheno genetic algorithmwith99-table. Numerical experiments illustrate that obtained model andalgorithm are effectiveness and feasibility.