Accurate modelling of lossy SIW resonators using a neural network residual kriging approach

In this paper, a computational intelligence method to model lossy substrate integrated waveguide (SIW) cavity resonators, the Neural Network Residual Kriging (NNRK) approach, is presented. Numerical results for the fundamental resonant frequency fr and related quality factor Qr computed for the case of lossy hexagonal SIW resonators demonstrate the NNRK superior estimation accuracy compared to that provided by the conventional Artificial Neural Networks (ANNs) models for these devices.