ON MODEL REDUCTION BASED ON WEIGHTED EQUATIONS ERRORS

The paper deals with the problem of approximating a stable continuous–time multivariable system by minimizing the L2norm of a weighted equation error. Necessary and sufficient conditions of optimality are presented, and some properties of the resulting approximating models, such as stability and uniqueness, are pointed out. Based on the conditions of optimality, an efficient algorithm for generating reduced–order models that retain different numbers of Markov parameters and MacLaurin expansion coefficients, is derived and applied to a benchmark example.