Pheromone Modification Strategies for Ant Algorithms Applied to Dynamic TSP

  • Authors:
  • Michael Guntsch;Martin Middendorf

  • Affiliations:
  • -;-

  • Venue:
  • Proceedings of the EvoWorkshops on Applications of Evolutionary Computing
  • Year:
  • 2001

Quantified Score

Hi-index 0.02

Visualization

Abstract

We investigate strategies for pheromone modification of ant algorithms in reaction to the insertion/deletion of a city of Traveling Salesperson Problem (TSP) instances. Three strategies for pheromone diversification through equalization of the pheromone values on the edges are proposed and compared. One strategy acts globally without consideration of the position of the inserted/deleted city. The other strategies perform pheromone modification only in the neighborhood of the inserted/deleted city, where neighborhood is defined differently for the two strategies. We furthermore evaluate different parameter settings for each of the strategies.