Effective robust optimization based on genetic algorithm

Although probabilistic optimization methods based on genetic algorithm (GA) provides accurate results, its performance is sometimes considerably sensitive to parameter changes. Moreover, the constraints are violated due to such parameter changes. A robust GA which performs random perturbation during optimization processes has been applied to some mathematical problems to show that it works as fast as the usual GAs. An adequate elite reservation technique for the robust GA is presented and applied to the robust GA for electromagnetic problems. Moreover, this method is shown to find solutions which are kept feasible against parameter changes.