An EM Optimization Method with Adjoint-Sensitivity-Based Multifeature Surrogate for the Design of Microwave Filters

This paper presents an efficient electromagnetic (EM) optimization method with adjoint-sensitivity-based multifeature surrogate for the design of microwave filters. This method is designed to solve the EM optimization problem where the starting point is far from the design specifications. Feature sensitivities are incorporated into the multifeatured surrogate modeling to achieve a surrogate with higher accuracy compared to the existing methods during the EM optimization process. A new design objective function in the feature space is also utilized to improve the speed of multifeature surrogate optimization. A microwave filter example is presented to illustrate this optimization method.