Abstract
In this paper a method for structured linear uncertain system reduction is presented. The reduced model is obtained by retaining [ beta super(-) sub(1), beta super(+) sub(1)] and [ alpha super(-) sub(1), alpha super(+) sub(1)] (for the steady state of reduced model and original system be equal) along with [ beta super(-) sub(i), beta super(+) sub(i)] and [ alpha super(-) sub(i), alpha super(+) sub(i)] for (i identical with 2, 3, ..., r - 1) which are contributing most to impulse response energy [I super(-) sub(s), I super(+) sub(s)]. A numerical example illustrates the procedure.