Fast parameter optimization using Kriging metamodeling [antenna EM modeling/simulation]

In this paper, we propose a hybrid optimizer, which combines the Kriging macro-modeling and the divided rectangles method (DIRECT) to perform global optimization. The latter yields a deterministic answer with fast convergence rate and possesses local as well as global optimization properties. Two examples are presented, where DIRECT optimizer is combined with Kriging metamodeling. The first example involves optimization of the shape of a slot array Frequency Selective Surface (FSS). In the second example, the electromagnetic coupling between the harness and the FM antenna on a Deville model automobile is minimized. Both examples demonstrate the exceptional rate of convergence and the flexibility of the proposed optimizer.