Bayesian identification of bolted-joint parameters using measured power spectral density

A Bayesian method for the optimal estimation of parameters that characterize a bolted joint based on measured power spectral density is proposed in this article. Due to uncertainties such as measurement noise and modelling errors, it is difficult to identify joint parameters of a bolted structure accurately with incomplete measured response data. In this article, using the Bayesian probability framework to describe the uncertainty of the joint parameters and using the power spectrum of the structural response of the single-point/multi-point excitation as measurements, the conditional probability density function of the joint parameters is established. Then, the Bayesian maximum posterior estimation is performed by an optimization method. Two simplified bolted-joint models are built in the numerical examples. First, the feasibility of the proposed method in the undamped model is proved. Then, taking advantage of multi-point excitation, the identification accuracy of the proposed method in the damped model is improved. The numerical results show that the proposed method can accurately identify the stiffness and damping characteristics of joint parameters with good robustness to noise. Finally, the joint parameters of the finite element model for an aero-engine casing are identified by the proposed method with satisfactory accuracy.