ORDER-REDUCED MODELS BASED ON TWO SIDES TECHNIQUES FOR INPUT-OUTPUT SYSTEMS GOVERNED BY DIFFERENTIAL-ALGEBRAIC EQUATIONS

摘要

In this paper, a new model order reduction method is presented for solving large-scale differential-algebraic equation (DAE) systems. By nonsingular matrix transforms, the large-scale DAE system is decomposed into an ordinary differential equation (ODE) subsystem and a DAE subsystem with the same index as the original system. A dual weighted H-2 model order reduction method is used to reduce the ODE subsystem, which can avoid the problem of large calculation caused by solving the Lyapunov equations. In order to keep the stability of the original DAE subsystem, we present a modified Lanczos model reduction (MLMR) method, which can produce a reduced-order model with better performances. Numerical experiments illustrate the effectiveness of our method.