| Summary: | 碩士 === 國立暨南國際大學 === 土木工程學系 === 95 === The Heterogeneous Fleet Vehicle Routing Problem（HVRP） is an important theoretical and practical topic in the study of routing problem. This study incorporates various local search methods and threshold values into genetic algorithm for solving a set of test instances adopted in the past studies. The performance of different local search methods and threshold values was investigated. The results showed that the single routing local search strategy perform well when the ratio between variable cost (i.e., cost of routing) and total cost gets higher. With high objective function value, the linear decreasing threshold value performs better than the fixed threshold value strategy. Moreover, the average obtained objective values can be improved 1.19% to 11.79% by adding four local search methods, which was developed in this thesis. Based on the best known solutions associated with different instances from the previous studies, the genetic local search method proposed in this study can obtain better objective function values in two instances and the same values in six instances among 12 testing instances.
|
|---|
