A note on the model reduction problem

A mixed method of model reduction is proposed; it is based on the differentiation method suggested by Gutman et al. [1] and on the usual Padé method. More precisely, the differentiation procedure is used to obtain the reduced-order denominator, thus ensuring stability preservation, while the numerator is obtained by matching the first MacLaurin expansion coefficients of the original system and model transfer functions. The method is computationally very simple and may lead to satisfactory results, as shown by an example where the original system transfer function is taken from the relevant literature.