A mathematical formulation for model reduction of discrete-time system
s such that the reduced-order model represents the system in a particu
lar frequency range is discussed. The algorithm transforms the full-or
der system into balanced coordinates using frequency-weighted discrete
controllability and observability grammians. In this form a criterion
is derived to guide truncation of states based on their contribution
to the frequency range of interest. Minimization of the criterion is a
ccomplished without need for numerical optimization. Balancing require
s the computation of discrete frequency-weighted grammians. Closed-for
m solutions for the computation of frequency-weighted grammians are de
veloped. Numerical examples are discussed to demonstrate the algorithm
.