Parameters Selection for Genetic Algorithms and Ant Colony Algorithms by Uniform Design

This paper presents a method based on Uniform Design (UD) to choose appropriate parameters of Genetic Algorithm (GA) and Ant Colony Algorithm (ACA) for solving Traveling Salesmen Problem (TSP). Although meta-heuristic algorithms, such as GA and ACA, are ideal optimizer for engineering practice, one of their drawbacks focuses on performance dependency on the parameters, which differ from one problem to another. It is interesting and significant to properly choose algorithms and their parameters to adapt different problems. Thus this paper utilizes UD to choose the limited representative parameter combinations and check the effectiveness with the traditional exploratory method. Experiments show the performance of GA and ACA with different parameter configuration and problems changes drastically.