Computationally-efficient Algorithm for Robust Optimization of Electromagnetic Design Problems

In the design stage of electromagnetic device, the way to handle the uncertainty of design parameters has attracted wide attention. Monte Carlo Method works well when dealing with uncertainties but it consumes too much time and computing resource. This paper proposes a computationally efficient way to achieve robustness based on Stochastic Collocation Method and TEAM 22 is used as a verification example. It is demonstrated that the approach combining Stochastic Collocation Method and genetic algorithm provides high computational efficiency without losing accuracy comparing with Monte Carlo Method.