A Self-adapting Immigrational Genetic Algorithm for Solving a Real-Life Application of Vehicle Routing Problem

Abstract

The present article presents the research for optimizing a real-life instance of heterogeneous vehicle routing problem, used intensively by shipping companies. The experiments have been carried out on the data provided by real companies, with constrains on the number and capacity of the vehicles, minimum and maximum number of stops for each route, along with the margins which can be take into account when optimizing the load of each truck. The optimization is performed using genetic algorithms hybridized with techniques for avoiding local optima, such as self adaptation and immigration. It turns out that more sophisticate approaches perform better, with very little compromise on execution time. It is a new proof of the importance of immigration techniques in bringing diversity in the genetic population.1.