Simulation of Complex High Order Systems using Reduced Order Models

A computer oriented algorithmic method is proposed to overcome the serious disadvantage of the classical Pade approximation technique often resulting in unstable reduced models from stable high order systems. For a desired kth order reduced model, the 2kth time moment is adjusted for stability while the initial (2k-1) time moments of the original system are retained in the reduced model. Some typical numerical examples show that this method gives a stable model with better time and frequency response matching properties.