Passivity-preserving model reduction for descriptor systems with H∞ norm error bounds

This paper considers the passivity-preserving model reduction problem for positive real descriptor systems. The model reduction principle is presented first, then a sufficient and necessary conditions is provided to check whether the high-order descriptor systems can be reduced with preservation of passivity. A model reduction algorithm is provided. Moreover, the presented method provides an H_∞ norm bound on the approximation error.