Multi-Objective Pareto Optimization of Electromagnetic Devices Exploiting Kriging With Lipschitzian Optimized Expected Improvement

This paper focuses on resolving the storage issue of correlation matrices generated by kriging surrogate models in the context of electromagnetic optimization problems with many design variables and multiple objectives. The suggested-improved kriging approach incorporating a direct algorithm is able to maintain memory requirements at a nearly constant level while offering high efficiency of searching for a global optimum. The feasibility and efficiency of this proposed methodology are demonstrated using an example of a classic two-variable analytic function and a new proposed benchmark TEAM multi-objective Pareto optimization problem.