SIMO system identification from measured ringdowns

Recent variations of Prony analysis and similar methods have proven to be valuable tools in identifying transfer functions and estimating the modal content from measured ringdowns. Such tools are becoming a standard in power system dynamic analysis. Current analysis methods assume the system to be single-input single-output (SISO) with distinct eigenvalues. Individual signals are analyzed independently often resulting in conflicting frequency and damping estimates (due to noise effects). Also, one cannot apply Prony analysis to systems with repeated or very closely-spaced poles such as in power system subsynchronous resonance problems. This paper presents a variation of Prony analysis that allows identification of a single-input multi-output (SIMO) proper model with possible repeated poles. Prony analysis as a signal analysis method is first generalized to handle multiple signals with poles that may be repeated. Expressions are then derived for the transfer function terms assuming the input is a series of square-wave pulses. Advantages of the SIMO analysis method are demonstrated on a power system Monte Carlo type simulation model. The example shows that the SIMO formulation allows for more accurate estimation of electromechanical oscillation modes under noisy conditions (such as field measured data).