Large Differential-Algebraic Systems with Parametric Uncertainty

Michael Lantz, Automatic Control, Lund Institute of Technology - Sweden

A general computational methodology is introduced for analysis of parametric uncertainty in large differential-algebraic systems. The objective is to convert the analysis problem into computation of structured singular values, while at the same time keeping the matrix dimensions down.

The proposed methodology has been implemented using the Omsim package for object-oriented modeling and Matlab for matrix computations. It is applied to a model of the Scandinavian power system, including 16 generators, 16 power loads and 20 transmission lines. There are 16 inputs, 16 outputs, 63 states and more than 500 parameters, of which 44 are assumed to be uncertain.