Antenna Modeling Based on Two-Stage Gaussian Process Considering Sensitivity Information

For antenna modeling and optimization design, high-fidelity full-wave electromagnetic simulation software can generally be used to obtain training samples. However, this process takes a long time. To solve problem, two-stage Gaussian (GP) considering sensitivity information is proposed. Based on the coarse grid fine of software, first stage learns mapping relationships performance each parameter, simultaneously. In second stage, accurate surrogates are established based above obtained. Because method in study into consideration during process, it better reflect relationship between input output, which develop more surrogate model. The proposed GP model significantly reduce demand samples for modeling, greatly save time calling software. Therefore, suitable problems insufficient approach evaluated by an inverted F ultra-wideband planar monopole antenna. results show that has high accuracy with limited given simulation, further verify its effectiveness efficiency.