A novel algorithm for optimum order estimation of nonlinear reduced macromodels

Estimation of an optimal order for reduced models is a challenging task and is often based on heuristics. In this paper, a new systematic algorithm is presented for estimating the minimum acceptable order for reduced models of nonlinear systems to ensure accurate and efficient transient behavior. The methodology incorporates the techniques developed in nonlinear time-series analysis, nonlinear model order reduction and computational geometry for a precise determination of the optimum order for a reduced nonlinear system.