MODEL-REDUCTION VIA PARAMETER MATCHING USING A GRAMIAN TECHNIQUE

Mode reduction technique based on parameter matching for single-input single-output (SISO) continuous systems is proposed. The reduced-order models match various combinations (chosen by the designer) of four ty pes of system invariant parameter of the full-order system associated with low-frequency response, high-frequency response, low-frequency po wer spectral density and high-frequency power spectral density. In thi s way, selected parameters of both transient and steady-state properti es can be preserved in the model reduction process. Which parameters a re chosen to be matched is contingent on those input/output properties of the original system that the designer considers most important to be retained in the reduced model. The number of parameters to be match ed determines the order of the reduced model. The effectiveness of the matched parameters in retaining the input/output properties of the or iginal system in the reduced models is fully investigated. The algorit hm presented finds not one but the entire set (infinite) of reduced mo dels that match a particular number of these parameters. Because the m ix of parameters desired (i.e. the algorithm chosen) is at the discret ion of the analyst, the new algorithm presented provides a handy assor tment of techniques instead of a single 'finished product'.