Treffer: Solving soft and hard-clustered vehicle routing problems: A bi-population collaborative memetic search approach.

The soft-clustered vehicle routing problem is a natural generalization of the classic capacitated vehicle routing problem, where the routing decision must respect the already taken clustering decisions. It is a relevant routing problem with numerous practical applications, such as packages or parcels delivery. Population-based evolutionary algorithms have already been adapted to solve this problem. However, they usually evolve a single population and suffer from early convergence especially for large instances, resulting in sub-optimal solutions. To maintain a high diversity so as to avoid premature convergence, this work proposes a bi-population collaborative memetic search method that adopts a bi-population structure to balance between exploration and exploitation, where two populations are evolved in a cooperative way. Starting from an initial population generated by a data-driven and knowledge-guided population initialization, two heterogeneous memetic searches are then performed by employing a pair of complementary crossovers (i.e., a multi-route edge assembly crossover and a group matching-based crossover) to generate offspring solutions, and a bilevel variable neighborhood search to explore the solution space at both cluster and customer levels. Once the two evolved new populations are obtained, a cooperative evolution mechanism is applied to obtain a new population. Extensive experiments on 404 benchmark instances show that the proposed algorithm significantly outperforms the current state-of-the-art algorithms. In particular, the proposed algorithm discovers new upper bounds for 16 out of the 26 large-sized benchmark instances, while matching the best-known solutions for the remaining 9 large-sized instances. Ablation experiments are conducted to verify the effectiveness of each key algorithmic module. Finally, the inherent generality of the proposed method is verified by applying it to the well-known (hard) clustered vehicle routing problem. • A collaborative bi-population memetic search is proposed for Soft and Hard CluVRP. • Two complementary crossovers are used for offspring generation. • A bilevel variable neighborhood search is used for cluster and customer-level optimization. • Improved best upper bounds are reported for 16 large instances. • Key factors to the good performance of the proposed algorithm are analyzed. [ABSTRACT FROM AUTHOR]

Copyright of European Journal of Operational Research is the property of Elsevier B.V. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)