Modular Neural Network for Coupled-Resonator Filter Modeling with EM-Sensitivity

This paper presents a study on the neural network-based surrogate modeling of coupled-resonator microwave filters. The proposed model consists of several modules that are constructed independently to form a modular neural network. To ensure the accuracy of the neural network model across a wide range of training data, electromagnetic (EM) sensitivity is exploited. To validate the effectiveness of the proposed model, a 6–2 dielectric filter design is tested.