Comparative Analysis among Evolutionary Optimization Methods: Genetic Algorithms, Particle Swarm Optimization and Differential Evolution

Since the emergence of Genetic Algorithms to solve optimization problems, a wide variety of evolutionary methods have been proposed and their applications have diversified across virtually all fields of knowledge. This paper presents a comparative analysis among three of the most widely applied evolutionary methods: Genetic Algorithms, Particle Swarm Optimization and Differential Evolution method. For this purpose, computer programs were developed for each method and applied to find solutions for two- and three-variable benchmark test functions. The results indicate that Differential Evolution Algorithm is the easiest to understand and to develop the programs; but it sometimes has difficulties to fulfilled the constrains and the results are not always accurate. Genetic Algorithms exhibits good approximation and carry out good with the constrains but the binary base make it to take a little more time and it can introduce some errors. The Particle Swarm Optimization method demonstrates good performance in both benchmark problems. However, this performance does not guarantee similar behavior across all types of optimization problems, as these methods heavily rely on empirically tuned control parameters and randomly generated initial populations