Artificial Neural Network Based RF-Model Generator-Version 0.2

摘要

This report describes the development, implementation, and usage of an Artificial Neural Network (ANN) based RF-model generator. Once the user has provided appropriate measurement or simulation data describing the behavior of an RF component, the model generator trains the ANN, resulting in an S-parameter or equivalent-circuit model for the RF component. The ANN-based RF-model generator has been implemented using the flexible input language of APLAC circuit simulation and design tool. Therefore, the existing optimization methods of APLAC can be directly utilized in ANN training. The final RF model obtained can be immediately used in any APLAC simulation.