This work presents a new optimization method for obtaining aggregated reduced models for a given single input and single output linear system by making use of an unknown row vector to construct an aggregation matrix. And all errors appearing in the process and the transfer functions of reduced models are analyzed and calculated. From minimizing these errors can the unknown vector be defined. That is, this method can optimize the reduced models endlessly. Even so, it is difficult to find the best answer because what is involved is a multivariable extreme-value problem with constraint condition....
