A Quasi-linearisation approach to trajectory based methods for nonlinear MOR

Model Order Reduction (MOR) techniques are means to find low dimensional approximations of large dynamical systems for applications in simulation and control. Trajectory based methods are a class of MOR techniques for nonlinear systems wherein multiple linearisations along a training trajectory are superposed to approximate the system dynamics. Although standard trajectory based methods generally work well, there are instances which require specific modifications in the approach. In this work, we propose a modification of the conventional technique for systems with nonlinear input operators. The modified method involves a quasi-linearisation of the nonlinear system and subsequent MOR and superposition. It is applied to a nonlinear circuit model for which conventional trajectory based methods have been found ineffective. The results obtained show a considerable improvement in the reproduction of the nonlinear system dynamics.