A Generalized Approach to Real-Time Performance Estimation of Antenna Types Using Deep Learning
Conference
Khan, MR, Zekios, CL, Bhardwaj, S et al. (2022). A Generalized Approach to Real-Time Performance Estimation of Antenna Types Using Deep Learning
. 497-498. 10.1109/AP-S/USNC-URSI47032.2022.9886506
Khan, MR, Zekios, CL, Bhardwaj, S et al. (2022). A Generalized Approach to Real-Time Performance Estimation of Antenna Types Using Deep Learning
. 497-498. 10.1109/AP-S/USNC-URSI47032.2022.9886506
Machine learning methods, such as deep learning, hold great potential in the study of electromagnetic (EM) structures as they excel at complex multi-dimensional modeling. In this work, we introduce a generalized method to predict antenna performance using deep neural networks. Specifically, we estimate the far-field radiation and S11 of an antenna (e.g., rectangular patch antenna) in real-time by utilizing the near-field data using an artificial neural network (ANN). Our method is generalized as it can learn from a range of design parameters for various substrate materials and operating frequencies, and it can accurately predict all the required EM properties (e.g., near-and far-fields, S-parameters) of the desired antenna. When compared to numerical methods, we achieved acceptable antenna performance estimation, e.g., a mean absolute percentage error of 0.27% for an arbitrary antenna configuration.
