Reduced-order modeling of transient power grid network with improved basis pod

Abstract

Power systems dynamics is represented by coupled equations and influenced by several elements. Most of these elements exhibit nonlinear behavior, and hence, computational efforts required for solving these coupled nonlinear equations are expensive. To simplify computational efforts, reduced-order model can be used. This reduced-order model is formulated by projecting a large-dimensional state vector onto a small-dimensional subspace spanned by an orthonormal basis which is conventionally obtained through a technique, called proper orthogonal decomposition (POD). However, this technique involves simulation of the high-dimensional nonlinear system to obtain the orthonormal basis for the projection. In this paper, we have presented an improvement of POD technique with the approximate snapshot ensemble bases extraction to avoid the need to simulate full-order dynamics of the large power grid network. The POD basis is generated from approximate trajectory which reduces simulation time with insignificant or no error in the employed model.