Threshold accepting metaheuristic with adaptive parameter setting for solving the vehicle routing problem with simultaneous pickups and deliveries

• Main Author: 趙致傑
• Other Authors: 韓復華
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/47508620800063608550

碩士 === 國立交通大學 === 運輸科技與管理學系 === 99 === The Vehicle Routing Problem with Simultaneous Pickups and Deliveries (VRPSPD) is an extension of the conventional VRP. Unlike the VRP which restricts the distribution for pickups or deliveries only, the VRPSPD is more flexible and applicable for solving real-world distribution problems in a supply chain. This paper proposes a new metaheuristic method for solving the VRPSPD based on the threshold accepting (TA) framework. A new randomize variable neighborhood decent (RVND) method, which incorporates 1-0, 1-1, 2-opt*, Or-Opt and 2-Opt local search methods, is applied in our solution method. In addition, we also propose an adaptive parameter setting system to help the planner determine appropriate parameters of initial threshold, end threshold and threshold reduced ratio value. Our proposed algorithms tested on a set of 72 benchmark instances including 14 instances from Salhi and Nagy, 18 instances from Tang Montané and Galvão and 40 instances from Dethloff. Results showed that our proposed methods have generated 40 best know solutions (BKS). The average deviation of all the tested instances is merely 0.42%.