A Surrogate-Assisted Hybrid Optimization Algorithms for Computational Expensive Problems

A surrogate-assisted hybrid optimization algorithms is proposed in this paper, in which trust region method is proposed to find an optimal position in the neighborhood for each individual so as to narrow the search space gradually, and particle swarm optimization is then utilized to find possible global optimum with the assistance of radial basis function network as the surrogate models. Empirical study on seven benchmark problems shows that the proposed method is capable of attaining high quality solutions under a limited computational budget.