Optimal model reduction with a frequency weighted extension (Q5929964)

This paper deals with model reduction by using an optimization technique. The objective function minimized is the impulse energy of the overall system. An extension of the technique to the frequency weighted case is also presented where single-sided or double-sided weightings can be incorporated in the reducing procedure. The authors propose an alternative to find an optimal solution by solving ordinary differential equations which are gradient flow associated with the objective function to be minimized. Two concrete examples are presented to illustrate the method. The numerical method is closely related to the gradient technique.