Artificial neural network modeling of component nonlinear behavior and application in conducted interference analysis

A method for determining nonlinear large-signal S-parameters from artificial neural network model trained with limited measured data is proposed. Predicted nonlinear S-parameters of a nonlinear device composed of a Schottky diode are in good agreements with measurements. Then the formula for calculating nonlinear S-parameters of network composed of a two-port nonlinear device cascaded with a three-port linear device is deduced, and the application of nonlinear S-parameters in conducted interference analysis is discussed. Finally, the applicability of the proposed method for conducted interference analysis involving nonlinear components is demonstrated by two different cascaded networks.