A memetic Algorithm for the Capacitated Vehicle Routing Problem

Abstract

In this paper we present a new memetic algorithm for the CVRP (Capacitated Vehicle Routing Problem). The new algorithm was developed from our earlier multi-objective algorithm for the vehicle routing problem - selecting and further developing one part of the earlier algorithm. The new algorithm is a steady-state evolutionary algorithm. It uses tournament selection; the descendents are derived from the parents by mutation based on the EVL (Extended Virtual Loser) where the EVL is an explicit collective memory technique. The algorithm is a memetic algorithm and uses five different stochastic 2-opt local searches to improve the descendents. We used some test problems of the Vehicle Routing Data Sets and of Christofides. Comparing the results with other method’s results we concluded: in the case of n < 200 costumers we got similar results that was published earlier.