An improved method using factor division algorithm for reducing the order of linear dynamical system.

An improved method is proposed to determine the reduced order model of large scale linear time invariant system. The dominant poles of the low order system are calculated by clustering method. The selection of pole to the cluster point is based on the contributions of each pole in redefining time moment and redefining Markov parameters. The coefficients of the numerator polynomial for reduced model are obtained using a factor division algorithm. This method is computationally efficient and keeps up the stability and input output characteristic of the original arrangement. [ABSTRACT FROM AUTHOR]