An age layered population structure genetic algorithm for the multi-depot vehicle problem

The vehicle routing problem is experienced in several industries such as transportation, distribution and logistics. We study the Multi-depot vehicle routing problem (MDVRP) a variant of the classical VRP with several depots and customers as a multi-objective problem where both the number of vehicles and total distance travelled are minimized simultaneously. A multipopulation structure, i.e., the age layered population structure genetic algorithms (ALPS-GA) with inter-layer transfer strategies is proposed. An empirical study is carried out using well-known MDVRP benchmark problems and using both the weighted sum and Pareto ranking fitness evaluation strategies. A comparative study with previously published genetic algorithms results shows the effectiveness of the proposed ALPS-GA approach.