シミュレーテッド・アニーリング法による配送計画問題の解法

概要

論文の詳細を見る
This paper is stated how the neighbouring solution of a present solution is generated for solving vehicle routing problems, in which many types of trucks starting from and returning to a depot will be toured for satisfying demands of cities. For solving the problems, two dynamic route generation methods are proposed ; the dynamic 2-change and the dynamic node insertion. But they are not completely the new methods. They are the modified route generations, which are originally stated as Lin's λ-opt and others. In these proposed methods, a new route is generated by linking two routes or two new routes are generated by dividing a route. These operations will be possible by adding a dummy city which has no demand, the same address as the depot and one directed link to the depot and viceversa. In simulated annealing method, they will be applied for generating shorter route sets as possible solutions in the solution space of the vehicle routing problems. The amount of demand of cities in the each route should be less than or equal to the capacity of the truck assigned. And as the cooling schedule of the annealing, Lundy and Mees's is employed in this paper. In the numerical experiments, the dynamic 2-change is well sensitive for generating the shorter route set than the other.
東海大学の論文