Bayesian identification of structural parameters of hybrid mid-frequency systems

A Bayesian technique is presented to evaluate the coupling characteristics of complex vibro-acoustic or structural dynamic systems. Although there is no restriction on the applicability of the method in the whole frequency range, any such coupled system may be considered to be a hybrid system in the sense that its components may behave in their low-, midor high-frequency regimes. Inexact measurements of the transfer functions, whose errors are assumed to be a random Gaussian process, are simulated. Based on some prior information and on the actual measurements, Bayes theorem then gives an expression of the probability density function of the coupling characteristics. This posterior probability is evaluated implicitly through sampling via a Markov chain Monte Carlo approach. The same approach can be used to evaluate other parameters of a complex system.