Optimization of Adaptive Genetic Algorithm Parameters in Traveling Salesman Problem

Abstract

The TSP problem is one where a seller visits multiple destinations at the same time and they are only allowed to visit once. The purpose of this TSP is to shorten the shortest distance, thereby minimizing time and cost. There are various methods to solve the TSP problem, including greedy algorithm, brute force algorithm, hill climbing method, ant algorithm, and genetic algorithm. Each process in a genetic algorithm is affected by several parameters, including population size, maximum generation, crossover rate, and mutation rate. The purpose of this study is to apply genetic algorithms to the traveling salesman problem optimization, calculate the maximum influence of generation, chromosome number, crossover rate and mutation rate on the optimal genetic algorithm, calculate the range of chromosome number, population size, crossover rate and mutation rate for genetic algorithm in the traveling salesman problem and calculate the effect of adaptive genetic algorithm parameters on genetic algorithm results. Based on the results obtained from research and testing, the four parameters of the genetic algorithm are positively correlated with fitness results while negatively correlated with execution time performance where each adaptive parameter applied provides more optimal fitness results than static parameters. The four adaptive parameters that are applied together give optimal results, both fitness which reaches 1.0% and time reaches 38.7%.