On the reliability of system identification: Applications of bootstrap theory

摘要

The estimation of structural damping from systems with unknown input can be particularlyrnchallenging but is vital to furthering the understanding energy dissipation in wind-inducedrnmotion. The Random Decrement Technique and Spectral Analysis are commonly invoked systemrnidentification schemes for this problem. Although there is well-established theory regarding thernbias and variance errors associated with these two approaches, the theoretical error formulae givernno indication of errors inherent to the estimated dynamic properties. This study utilizes bootstrappedrnreplicates of the random decrement segments and raw power spectra to assess the qualityrnof the system identification by providing surrogate estimates of damping and natural frequency torngenerate useful statistics and confidence intervals.