High Dimensional Modeling of Microstrip Hairpin Bandpass Filter Using Artificial Neural Networks

Abstract

Conventional neural network modeling techniques are not suitable for developing models that have many input variables because data generation and model training become too expensive. In this paper, an efficient neural network modeling technique for microstrip hairpin band pass filter that have many input variables is proposed. The decomposition approach is used to simplify the overall high dimensional neural network modeling problem into a set of low dimensional sub neural network problems. A method to combine the sub models with a filter empirical/equivalent model is developed. An additional neural network mapping model is formulated with the neural network sub models and empirical/ equivalent model to produce the final overall filter model. Even, with a limited amount of data, the proposed model can produce much more accurate results compared to the conventional neural network model and the resulting model is much faster than an EM model.