Un estudio comparativo sobre evolución diferencial auto-adaptativa en ambientes dinámicos A comparative study on self-adaptive differential evolution in dynamic environments

Several real optimization problems are dynamic, meaning that some elements of their mathematical model are time varying. These problems have received a special interest in the last years from the viewpoint of metaheuristics. Differential Evolution (DE) is one of the current population-based metaheuristics with an excellent effectiveness and easy implementation. However, as similar paradigms in dynamic environments, DE has been adapted with aims of solving the diversity loss in the solution population. On the other hand, self-adaptation is one of the less used approaches in dynamic environments, despite its success in complex scenarios. Self-adaptation is a parameter control technique that gives certain intelligent behavior to the algorithm, during the search process. In that sense, the present work investigates the performance of two self-adaptive extensions of DE, which in combination with other existing diversity approaches have been applied in several dynamic scenarios. The obtained results from the computational experiments, confirm that self-adaptation is a promising technique for dynamic environments.