作　　者：Lim, Andrew;Zhu, Wenbin

作者单位：[1]Hong Kong Univ Sci & Technol, Dept Ind Engn & Logist Management, Kowloon, Hong Kong, Peoples R China;[2]S China Univ Technol, Sch Engn & Comp Sci, Guangdong, Peoples R China

会议名称：ADVANCES IN APPLIED ARTICIAL INTELLIGENCE, PROCEEDINGS

摘　　要：Multi-depot Vehicle Routing Problem has great practical value to the logistics and supply chain management. The fixed distribution of vehicles variant (MDVRPFD) brings it one step closer to the practical use. Based on the simple fact that all sub-routes of an optimal route must be optimal, a new randomized best insertion (RBI) algorithm is proposed. The proposed insertion algorithm is highly effective in minimizing number of vehicles and fast. Compared to the best known result published for MDVRPFD variant, the solutions generated by this new insertion algorihtm require 20% fewer vehicles. Adopting a generalized n-op neighborhood operator, a Simulated Annealing approach yields a reduction of 12% in total distance compared to best known results of MDVRPFD.

关 键 词：MDVRP   MDVRPFD   meta-heuristic   insertion algorithm   sub-route optimization