| Popis: |
Vehicle routing is a logistical combinatorics optimization problem where one attempts to find the best way to send out a fleet of delivery trucks. Evolutionary algorithms have been developed to solve this problem in the past and have demonstrated a strong efficacy with respect to organizing the route that each truck takes. In this paper, a novel representation which makes use of the adaptive generative representation for an evolutionary algorithm is presented; the representation achieves marginally better results at organizing efficient routes than the original adaptive generative representation for certain groupings of circumstances and performs competitively with it in the remainder of circumstances. The novel representation is a modification to the way that the fitness function interprets the chromosome, requiring no other algorithmic modifications. Rather than encoding the next destination in order along the chromosome, chromosome loci are tied to destinations and encode choices of next destinations from each loci. |
