Application of Ordered Fuzzy Numbers in a New OFNAnt Algorithm Based on Ant Colony Optimization

This paper describes the results of experiments concerning the optimization method called OFNAnt. As the benchmarks, the author used the set of files from TSPlib repository which includes well known samples of the travelling salesman problem. The innovation of the proposed method consists in implementation of Ordered Fuzzy Numbers to the decision-making process of individual ant agents. This also made it possible to correlate the colony development optimization with the trend. Previous implementations of the fuzzy logic to such meta heuristics like ant systems came down to fuzzy control over the decision-making process of an ant or fuzzy control of the pheromone release mechanism. Thanks to the proposed method, it was possible to expand the family of solutions with the solutions represented by ants moving outside the main circulation. The improvement was possible thanks to better stress that, according to the OFN arithmetic, was put on their participation in the process as compared to the conventional approach, as the direction of their movement has been opposed to the trend followed by majority of colonies. Final conclusions of the experiment indicate to superiority of methods based on Ant Colony Optimization, and in particular the superiority of OFNAnt method over heuristic methods.