RBF metamodel assisted global optimization method using particle swarm evolution and fuzzy clustering for sequential sampling

To enhance the efficiency of modern engineering optimization problems involving computationally intensive analysis models, a new radial basis function (RBF) metamodel based global optimization method using particle swarm evolution and fuzzy c-means clustering, notated as PSFC-RBF, is presented. In PSFC-RBF, particle swarm evolution is used to generate a large amount of inexpensive samples, and then fuzzy c-means clustering is used to cluster the cheap samples for identifying the interesting space. Sequential expensive samples are produced to update RBF metamodel and lead the optimization process converge to the global optimum in an efficiency manner. PSFC-RBF is validated by using several numerical benchmark problems and an engineering problem. In terms of the comparison results with some other metamodel-based optimization methods, PSFC-RBF shows satisfactory performance in both optimization efficiency and global convergence capability. Moreover, the good robustness of PSFC-RBF is also demonstrated.