Modeling and Prediction of Integrated Passive Device Filter Based on Artificial Neural Network

This paper presents a modeling and prediction method of integrated passive device (IPD) filters based on the artificial neural network (ANN) model. The ANN forward model is trained to directly generate the distribution of poles and residues of the filter under a group of geometric parameters, while the ANN inverse model is established to generate the geometric dimensions with the poles and residues. The effectiveness of the proposed method is verified by the comparison among the ANN predicted, circuit and full-wave simulated frequency responses. The design of such an IPD filter can be simplified significantly.