An Unsupervised Microwave Filter Design Optimization Method Based on a Hybrid Surrogate Model-Assisted Evolutionary Algorithm

In resonator-coupled bandpass filter 3D design, it is a routine that the optimization methods are guided/supervised by designers’ experience to carry out an iterative design process. To realize automated or unsupervised optimization, new method, called hybrid surrogate model-assisted evolutionary algorithm for (H-SMEAFO), proposed. H-SMEAFO aims automatically obtain highly optimal without interaction (i.e., unsupervised) and also not restricted certain kinds of structures. H-SMEAFO, key innovations include response feature-based objective function global algorithm; both designed bespoke landscape characteristics. The performance demonstrated 8th-order dual-band waveguide with four transmission zeros 6th-order two zeros, which, does appear be possible using existing methods. Numerical results show effectiveness advantages H-SMEAFO.