A transformation approach for model order reduction of nonlinear systems

A novel technique for nonlinear model order reduction is developed. A transformation matrix is devised from a representative model of the original nonlinear system and applied to the nonlinear model to provide a reduced-order nonlinear model. Linearization is not necessary. Most methods used to compute the transformation matrix for linear system model reduction, such as balancing and aggregation, can be used. In certain situations, the linear system obtained by linearizing the nonlinear system about its equilibrium point can be used as the representative model. To illustrate the proposed procedure, a numerical example is included