A New Model Reduction Method for the Linear Dynamic Systems and Its Application for the Design of Compensator

A new model reduction method for the simplification and design of a controller for the linear time-invariant systems is proposed. An improved generalized pole clustering algorithm is employed in the proposed technique for obtaining the denominator of the reduced model. The numerator is computed with a simple mathematical technique available in the literature. The proposed method guarantees the stability in the reduced plant given that the full-order plant is stable and also retains the fundamental characteristics of the original model in the approximated one. This reduced model has been used for the design of compensator for the large-scale original plant by using a new algorithm. The compensator obtained by using the reduced model gives the approximately same time domain specification as compensator obtained by using large-scale original system, and the design of compensator by using the reduced model is comparatively easier. The results of the proposed algorithm are compared with existing methods of reduced-order modeling which show improvement in the performance error indices, time response characteristics and time domain specifications. The validity, effectiveness and superiority of the proposed technique have been demonstrated through some standard numerical examples.