Fminsearch Optimization Based Model Order Reduction

A hybrid approach for model order reduction is proposed in this paper. The approximants for denominator polynomial are derived by matching both Markov parameters and Time moments, whereas numerator polynomial derivation and error minimization is done using Fminsearch algorithm. The efficiency of the proposed method can be investigated in terms of closeness of the response of reduced order model with respect to that of higher order original model and a comparison of the integral square error as well.