용량제한을 갖는 차량경로 문제에 대한 메타 휴리스틱의 비교 분석

Abstract

Most of the research on the vehicle route problem is about improved neighboring search method because it can save the time required to repeat the search process by efficiently constructing good neighbors. However, when comparing and evaluating the results of the proposed algorithm, there is a limit to evaluating it based on one meta heuristic. Therefore, in this paper, rather than presenting a new algorithm for the vehicle route problem, GLS, TSA, SA and greedy descent algorithm(GDA), which are representative meta-heuristics widely used to solve various types of vehicle route problems, are applied to CVRP problems and compared and analyzed. Through this, we tried to compare and evaluate the characteristics of four representative meta-heuristic techniques. For this purpose, the comparison and analysis between the techniques were applied to the CVRP benchmark problems known to be optimal. Even when the number of vehicles is large or at least the service capacity of the vehicle is variable and the number of demands increases, the guided local search (GLS) provides the best route in most situations. Another interesting thing about the experiments in this paper is that SA's solutions present the same results as the results of GDA in every model tested.