Bayesian operational modal analysis with asynchronous data, Part II: Posterior uncertainty

- Analytical expressions for efficient calculation of identification uncertainty.
- Examples with synthetic, laboratory and field data.
- Parametric study investigating the effect of modelling error.

A Bayesian modal identification method has been proposed in the companion paper that allows the most probable values of modal parameters to be determined using asynchronous ambient vibration data. This paper investigates the identification uncertainty of modal parameters in terms of their posterior covariance matrix. Computational issues are addressed. Analytical expressions are derived to allow the posterior covariance matrix to be evaluated accurately and efficiently. Synthetic, laboratory and field data examples are presented to verify the consistency, investigate potential modelling error and demonstrate practical applications.