Discussion on: 'Generalized LFT-Based Representation of Parametric Uncertain Models'

摘要

The paper by Hecker and Varga presents a generalized LFT-representation that often leads to lower order models than those obtained using standard LFT-representations. In this note it was shown that any linear fractional transformation relating the uncertain parameters to the normalized parameters can be used for normalization. It is important to note that normalization must be performed as late as possible in the realization procedure and must be followed by model reduction. The extension of multidimensional model reduction techniques to descriptor systems would be a valuable addition to this approach.