STRUCTURAL DAMPING

The following study investigates the reliability of Random Decrement Signatures in the presence of nonstationarity and when correlation is allowed between samples averaged in the signature. The effect of such features upon signature variance was not investigated previously and was facilitated through the use of bootstrapping theory. A limiting case error is proposed in light of the findings of this study to permit variance estimates when correlations between samples are permitted, greatly increasing the number of samples available for averaging. In general, the effects of correlation are marginal, while the two classes of non-stationarity considered produce significant deviations in error estimates from theory.