Modified DE Algorithms for Solving Multi-depot Vehicle Routing Problem with Multiple Pickup and Delivery Requests

Abstract

This chapter proposes two DE algorithms for solving multi-depot vehicle routing problem (MDVRP) with multiple pickup and delivery requests (GVRP-MDMPDR). Two modified DE algorithms based on subgrouping of vectors and strategy switching concepts are developed and evaluated. In the first proposed algorithm, subgrouping of vectors is applied in the crossover process of the classical version of DE algorithm. The vectors are divided into two subgroups. The first subgroup applies exponential crossover process, while the other subgroup applies binomial crossover process. Experiences from two different crossover approaches are shared by allowing a target vector to randomly select vectors from both groups during the mutation process. The other algorithm is based on strategy switching concept applied in the crossover process. Two different crossover processes, exponential and binomial crossover processes, are used alternately. The results obtained from these two proposed algorithms are compared to those obtained by the classical DE algorithm. The results show that both proposed DE-based algorithms outperform the classic DE.