Design of microwave devices by segmentation, finite elements, reduced-order models, and neural networks

An accurate and fast neural model for complex microwave circuits is efficiently obtained by using segmentation and exploiting the knowledge of frequency response obtained from reduced order models. Information arriving from the excited modes in the connection ports of the regions to be modeled is included analytically. The frequency response modeling is similar to that of reduced order Padé models, which are used to compute the necessary training data easily. Overall optimization can then be used efficiently. Several examples that illustrate the capacities of the method are also presented. © 2007 Wiley Periodicals, Inc. Microwave Opt Technol Lett 49: 655–659, 2007; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/mop.22448