Parameter Optimization in GA for Large-scale Traveling Salesman Problem

Abstract

Traveling Salesman Problems (TSP) is a typical problem in computer science and it is also the most concerned problem for researchers in combinatorial optimization. Therefore, how to solve the optimal loop of TSP problem is a hot topic in researchers. At present, the genetic algorithm has been mature in solving the TSP problem, but the research on the optimization of genetic algorithm parameters is still relatively less. Therefore, in this paper, we analyzed the relevant parameters of the genetic algorithm, optimized the parameters in the application of TSP problem, strengthened its optimization ability, and gave a calculation framework: Firstly, we sampled and analyzed the parameters by using the experimental design method and found the most influential parameters. Secondly, we established and optimized the proxy model and obtained the optimal parameters. Finally, we verified the results with examples. The experiment proved that this method improved the efficiency of solving TSP problem greatly.