Genetic Algorithm for Optimizing Traveling Salesman Problems with Time Windows (TSP-TW)

Abstract

The concept of Traveling Salesman Problem (TSP) used in the discussion of this paper is the Traveling Salesman Problem with Time Windows (TSP-TW), where the time variable considered is the time of availability of attractions for tourists to visit. The algorithm used for optimizing the solution of Traveling Salesman Problem with Time Windows (TSP-TW) is a genetic algorithm. The search for a solution for determining the best route begins with the formation of an initial population that contains a collection of individuals. Each individual has a combination of different tourist sequence. Then it is processed by genetic operators, namely crossover with Partially Mapped Crossover (PMX) method, mutation using reciprocal exchange method, and selection using ranked-based fitness method. The research method used is GRAPPLE. Based on tests conducted, the optimal generation size results obtained in solving the TSP-TW problem on the tourist route in the Province of DIY using genetic algorithms is 700, population size is 40, and the combination of crossover rate and mutation rate is 0.70 and 0.30 There is a tolerance time of 5 seconds between the process of requesting distance and travel time and the process of forming a tourist route for the genetic algorithm process.