Canonical modeling of nonlinear devices through circuit optimization

The author presents an approach to the canonical modeling of nonlinear resistive devices using a numerical optimization method. On the basis of the given measured data a nonlinear circuit model of optimally chosen elements is generated. Several types of nonlinear circuits corresponding to the appropriate approximation method are considered: piecewise-linear, polynomial, rational, and piecewise polynomial. Two approaches of hessian estimation have been considered: one with an analytical (exact) derived expression and the second applicable in more general circuits based on the Gauss-Newton approximation with Marquardt regularization. The illustrative examples obtained using the Newton method confirm the efficiency of the proposed solution