Design of Broadband Microwave Metamaterial Absorber using Surrogate-based Constrained Dual-objective Optimization

Optimization of a broadband metamaterial absorber usually adopts evolutionary algorithms, which require many objective function evaluations. To remedy this issue, a dual-objective surrogate-based constrained optimization algorithm is proposed to increase the absorption bandwidth and reduce the thickness of the absorber. To decrease the number of expensive evaluations, Kriging models are used to approximate objective functions. Then, a dual-objective expected improvement infill sampling criterion and probability of feasibility sampling criterion are considered to find current optimal solution and update the surrogate model. The optimal trade-off is obtained by Pareto front technique. The proposed algorithm is compared with a well-known multi-objective evolutionary algorithm NSGA-II to verify its feasibility and efficiency. Finally, a broadband metamaterial absorber is optimized with the absorption bandwidth from 9.77GHz to 32.21GHz and the thickness of 2.93mm.