Surface Wave Antenna Metallic Cell Pattern Design Using Neural Network Method

This work presents a surface wave antenna metallic cell pattern prediction method which can be generated based on the required far-field radiation pattern by the mean of applying Wasserstein generative adversarial network (WGAN) and bi-directional gated recurrent unit (Bi-GRU) neural network models. The predicted metallic cell pattern has been 3D-modelled in CST and the radiation pattern shows less than 1 dBi variation level from the desired input radiation pattern.