Location Modeling
Mona Alizadeh Firozi; Vahid Kiani; Hossein Karimi
Abstract
Purpose: The purpose of this paper is to propose an improved genetic algorithm to solve the problem of Uncapacitated Single-allocation Hub Location. Previous methods have paid less attention to the diversity of population, and due to insufficient vairation in mutation operators, they perform well only ...
Read More
Purpose: The purpose of this paper is to propose an improved genetic algorithm to solve the problem of Uncapacitated Single-allocation Hub Location. Previous methods have paid less attention to the diversity of population, and due to insufficient vairation in mutation operators, they perform well only in a few runs, and in other runs they are caught in the local optimum.Methodology: The proposed method uses appropriate genetic operators to increase diversity of the population and performs local search around the best answer to exploit promising areas of the solution space. The use of hub mutation operators along with allocation mutation operators in the proposed algorithm has increased its exploration ability and effectiveness, which has led to discovery of the optimal answer in most runs for large size problems. Also, searching for the local neighborhood of the best answer made convergence faster and reduced the total running time for large instances.Findings: Evaluation of the proposed method and base algorithm on the Australian Post (AP) dataset showed that the improvements increased efficiency of the genetic algorithm in achieving optimal solutions for problems as large as 200 nodes from 2% to more than 85%.Originality/Value: This study showed that meta-heuristic algorithms and their improved versions are suitable methods for solving hub location problems in a short and limited time.