Model reduction of discrete-time systems via power series transformation

Presents a new algorithm for model reduction of z-transfer functions. The simplification technique is based on moment matching. However, the simple algorithm is obtained with the use of a power-series transformation. Reduced-order models are obtained by matching coefficients of the power series in the transformed domain, which is equivalent to matching of time moments of impulse responses. Although it does not guarantee stability for the reduced transfer function the proposed technique produces good results, i.e. the difference between the step responses of the original and simplified systems is very small and diminishes rapidly