Optimization of the Ant Colony System Algorithm to Search for Distance and Shortest Routes on Travel Salesman Problems

Abstract

: The travel salesman problem is a combinatorial optimization problem that is very well-known in graph theory. The travel salesman problem is categorized as a difficult problem when viewed from a computational point of view. Also includes the classic "NP-Complete" problem because it has been studied for decades. TSP can be viewed as a matter of finding the shortest route that must be taken by someone who departs from his hometown to visit each city exactly once and then returns to his hometown of departure. In the travel salesman problem, the colony can coordinate through a very simple interaction, through this interaction, the colony is known to be able to solve very difficult problems. So, the method used to solve this TSP problem, using the Ant System algorithm is modified to the Ant Colony System Algorithm, to improve its performance on larger TSP problems. The main principle used in the AS algorithm is still used in the Ant Colony System algorithm, namely the use of positive feedback through the use of pheromones. A pheromone placed along the route is intended, so that the ants are more interested in taking that route. So that the best solution later, has a high concentration of pheromones. In order not to get trapped in the local optimal, negative feedback is used in the form of pheromone evaporation. While the main differences between the Ant System and Ant Colony System algorithms are different state transition rules, different global pheromone renewal rules, and the addition of local pheromone renewal rules. With this modification, the optimization results on the TSP obtained will be better, and get the shortest route in the minimum possible time. Based on the results of the system trials that have been carried out, it shows that the ant algorithm, both Ant Colony System and Ant System can be applied to the Travel Salesmen Problem. The Ant Colony System algorithm still has a faster search time than the Ant System algorithm and the difference is quite large.