Lumped model identification based on a double multi-valued neural network and frequency response analysis

Abstract

A novel identification technique for lumped models of general distributed circuits is presented. The approach is based on two multi-valued neuron neural networks used in a joined architecture able to extract hidden parameters, whose convergence allows the validation of the approximated lumped model. The inputs of the neural network are geometrical parameters of a given structure, while the outputs represent the estimation of the lumped circuit parameters. The method uses a Frequency Response Analysis (FRA) approach in order to elaborate the data to present to the net.