Accurate EC-ANN modeling for a RF-MEMS extended tuning range varactor

A novel accurate and efficient modeling method based on Equivalent Circuit trained Artificial Neural Network (EC-ANN) technique is developed for a RF-MEMS extended tuning range varactor. The parameters are extracted directly from the equivalent circuit model and used as training and testing sets for the ANN. Experiments show that the proposed approach can be used to fast and accurately model the RF characteristics of the RF-MEMS varactor. The results can agree with the EC-ANN predictions and the Ansoft HFSS simulations. To extend the capabilities of the proposed methodology, the developed EC-ANN modeling technique is used for design, simulation and optimization of the MEMS circuits.