Artificial Immune Algorithm for Multi-Depot Vehicle Scheduling Problems
Zhongyi Wu,Donggen Wang,Linyuan Xia,Xiaoling Chen
DOI: https://doi.org/10.1117/12.812864
2008-01-01
Abstract:In the fast-developing logistics and supply chain management fields, one of the key problems in the decision support system is that how to arrange, for a lot of customers and suppliers, the supplier-to-customer assignment and produce a detailed supply schedule under a set of constraints. Solutions to the multi-depot vehicle scheduling problems (MDVRP) help in solving this problem in case of transportation applications. The objective of the MDVSP is to minimize the total distance covered by all vehicles, which can be considered as delivery costs or time consumption. The MDVSP is one of nondeterministic polynomial-time hard (NP-hard) problem which cannot be solved to optimality within polynomial bounded computational time. Many different approaches have been developed to tackle MDVSP, such as exact algorithm (EA), one-stage approach (OSA), two-phase heuristic method (TPHM), tabu search algorithm (TSA), genetic algorithm (GA) and hierarchical multiplex structure (HIMS). Most of the methods mentioned above are time consuming and have high risk to result in local optimum. In this paper, a new search algorithm is proposed to solve MDVSP based on Artificial Immune Systems (AIS), which are inspirited by vertebrate immune systems. The proposed AIS algorithm is tested with 30 customers and 6 vehicles located in 3 depots. Experimental results show that the artificial immune system algorithm is an effective and efficient method for solving MDVSP problems.