A memetic approach to vehicle routing problem with dynamic requests

The paper presents an effective algorithm for solving Vehicle Routing Problem with Dynamic Requests based on memetic algorithms. The proposed method is applied to a widely-used set of 21 benchmark problems yielding 14 new best-know results when using the same numbers of fitness function evaluations as the comparative methods. Apart from encouraging numerical outcomes, the main contribution of the paper is investigation into the importance of the so-called starting delay parameter, whose appropriate selection has a crucial impact on the quality of results. Another key factor in accomplishing high quality results is attributed to the proposed effective mechanism of knowledge transfer between partial solutions developed in consecutive time slices. While particular problem encoding and memetic local optimization scheme were already presented in the literature, the novelty of this work lies in their innovative combination into one synergetic system as well as their application to a different problem than in the original works. (C) 2016 Elsevier B.V. All rights reserved.