Parametric Modeling of SISL Fourth-Order Coupled Line Bandpass Filter Using Neuro-TF Method

The substrate integrated suspended line (SISL) is a high-frequency transmission platform, which has many excellent characteristics such as low loss and low relative dielectric constant when transmitting signals in the microwave frequency band. However, due to the complexity of its structure, electromagnetic simulation of SISL is often time-consuming. Artificial neural network has an excellent performance in nonlinear relationship mapping. In this paper, SISL components are modeled by neural networks and transfer function using pole-residue format, namely neuro-transfer function (neuro-TF). The issue of discontinuity of coefficients is solved by order-changing algorithm. Then a SISL parallel coupled line bandpass filter is used to validate the model.