Parameter Setting in a Guided Local Search Metaheuristic for the Capacitated Vehicle Routing Problem
Guided Local Search (GLS) metaheuristic is a general optimization technique suitable for combinatorial optimization problems. The method works by adding penalty factor to the objective function, especially features that have high cost. GLS uses a mechanism to select features to penalize. We are goi...
| Main Author: | |
|---|---|
| Format: | Dissertation (University of Nottingham only) |
| Language: | English |
| Published: |
2009
|
| Online Access: | https://eprints.nottingham.ac.uk/22814/ |
| Summary: | Guided Local Search (GLS) metaheuristic is a general optimization technique suitable for combinatorial optimization problems. The method works by adding penalty factor to the objective function, especially features that have high cost. GLS uses a mechanism to select features to penalize. We are going to perform computational experiments with the GLS metaheuristic for the Capacitated Vehicle Routing Problems (CVRP). Given a set of customers i requiring a visit with associated demand qi, CVRP aims to construct a set of routes for m vehicles of identical capacity Q that minimizes the cost of operation. In performing our experiment, we will intentionally make changes to the input parameters in order to observe corresponding changes in the quality of obtained solutions. The objective of the experiments is to investigate several aspects related to the parameter setting of the penalization in the GLS. Based on the results, we construct a guideline of how to set the parameters for an effective GLS run. A new multiphase strategy is proposed in this dissertation. |
|---|