Stochastic model updating for assembled structures with bolted joints using a Bayesian method

An efficient model-updating method based on Bayesian power spectral sensitivity analysis is proposed to update the uncertain parameters of assembled structures. The dynamic equations of bolted assembled structures are derived using the substructure component model synthesis technique. The posteriori probability density function of uncertain parameters of bolted joints is established by the Bayesian method, where the negative logarithmic likelihood function is taken as the objective function to be optimized. To improve the efficiency of the stochastic model-updating process, the pseudo-excitation method is introduced to derive analytically the expressions of the gradient vector and the Hessian matrix for optimization. The proposed method is evaluated in the stochastic model updating of an assembled structure consisting of three beams. Then, it is applied to model updating of the simplified model of a rocket. The numerical results demonstrate that this approach can significantly reduce the computational cost and ensure computational accuracy