Effect of Reduced Model Order on Accuracy of Trajectory Piecewise Linear Approximations for a Class of Nonlinear Circuits

Abstract

Model order reduction of input affine nonlinear systems via trajectory piecewise linear approximation is a well known practice. This method along with its variants is known to generate efficient and accurate reduced models of large order nonlinear systems. The selection of the order of the reduced model is however a heuristic choice that comes with experience. There is no concrete measure of an optimum selection of the reduced order that would lead to approximations with high accuracy and least computational cost. This paper provides a study of few variants of trajectory piecewise linear method and effect of the choice of reduced order on the accuracy of the approximations and the computational cost. The results have been studied on a nonlinear transmission line circuit. This study provides a performance analysis that exhibits a range of acceptable values of reduced order that may be preferred for generating accurate trajectory piecewise approximations of circuits similar to that discussed in this paper.