Model Order Reduction for Networks网络模型降价.pdfVIP

Model Order Reduction for Networks of ODE and PDE Systems Michael Hinze and Ulrich Matthes Department of Mathematics, University of Hamburg, Bundesstr. 55, 20146 Hamburg, Germany michael.hinze@uni-hamburg.de ulrich.matthes@math.uni-hamburg.de Abstract. We propose a model order reduction (MOR) approach for networks containing simple and complex components. Simple compo- nents are modeled by linear ODE (and/or DAE) systems, while complex components are modeled by nonlinear PDE (and/or PDAE) systems. These systems are coupled through the network topology using the Kirch- hoff laws. As application we consider MOR for electrical networks, where semiconductors form the complex components which are modeled by the transient drift-diffusion equations (DDEs). We sketch how proper orthog- onal decomposition (POD) combined with discrete empirical interpola- tion (DEIM) and passivity-preserving balanced truncation methods for electrical circuits (PABTEC) can be used to reduce the dimension of the model. Furthermore we investigate residual-based sampling to construct reduced order models which are valid over a certain parameter range. Keywords: Model Order Reduction, Parametrized Dynamical Systems, Drift-Diffusion Equations, Integrated Circuits. AMS subject classifications: 93A30, 65B99, 65M60, 65M20. 1 Introduction In this paper we propose a simulation-based MOR approach for the reduction of networks consisting of (many) simple and (only few) complex components. We assume that the simple and complex components are modeled by systems of linear ODEs (DAEs) and nonlinear PDEs (PDA