Bayesian support vector regression (BSVR) modeling ofplanar antennas with reduced training sets for computational efficiencyis presented. Coarse-discretization electromagnetic (EM) simulations areexploited in order to find a reduced number of fine-discretization trainingpoints for establishing a high-fidelity BSVR model of the antenna. Asdemonstrated using three planar antennas with different response types,the proposed technique allows substantial reduction (up to 48%) of thecomputational effort necessary to set up the fine-discretization trainingdata sets for the high-fidelity models with negligible loss in predictivepower. The accuracy of the reduced-data BSVR models is confirmed bytheir successful use within a space mapping optimization/design algorithm.
展开▼
