Full Length Research Paper

Population pre-selection operators used for generating a non-random initial population to solve vehicle routing problem with time windows

Ocotlán Díaz-Parra, Jorge A. Ruiz-Vanoye* and José C. Zavala-Díaz

Universidad Autónoma del Estado de Morelos - FCAeI, Mexico.
 

*Corresponding author. E-mail: jorge@ruizvanoye.com.

Accepted 13 October, 2010

   Abstract

In this paper, we propose two population pre-selection operators to be improved by the k-means algorithm (clustering) and neighborhood techniques in the determination of the initial population (non-random population) used in the evolutionary and genetic algorithms to solve the vehicle routing problem with time windows (VRPTW). The results provide individual or group chromosomes in the initial population that is close to the optimal in VRPTW instances.

Key words: Pre-selection operator, k-means algorithm, neighborhood, logistics, vehicle routing, VRPTW.