Model reduction of large-scale interconnected systems

The paper addresses model reduction of large-scale interconnected systems in the component connection model framework. It is shown that model reduction at the component level is equivalent to model reduction at the system level. A general procedure is presented to obtain a reduced-order model when (a) at least one of the components is exactly aggregable, and (b) none of the components are reducible. In the latter case, a reduced-order model is obtained by keeping the key state variables and compensating for the deleted ones using a numerical QL algorithm. An interconnected electric power system example is given to illustrate the proposed procedure.