The high-stability bracket is the primary component in remote sensing satellites
responsible for carrying the camera and attitude control equipment; its dynamic
behavior directly affects imaging quality and in-orbit service life.
Welding-based connections introduce uncertainties during manufacturing and
assembly, leading to significant discrepancies between finite element analysis
results and experimental data. To quantitatively assess the impact of welding
processes on the accuracy of the high-stability bracket’s finite element
model(FEM), this study proposes a Bayesian parameter correction method based on
Markov Chain Monte Carlo (MCMC) sampling and the Metropolis-Hastings (MH)
algorithm. It employs a multilevel correction experimental
framework—”material-level, component-level, and system-level”—to progressively
refine material parameters and establish a high-fidelity FEM. First, the elastic
modulus of the welded and non-welded regions were calculated using stress-strain
curves obtained from static tensile tests on welded and non-welded plates; then,
targeting the first two natural frequencies from dynamic impact tests on bar
specimens, the material densities of the welded and non-welded regions were
corrected using the Bayesian method; finally, the effectiveness of the
correction method was verified through free vibration tests on the actual
high-stability bracket. The results indicate that, compared to the undamaged
region, the welded region exhibits a 13.87% loss in elastic modulus and a 55.32%
loss in density. The relative errors in the first three natural frequencies of
the corrected model were 0.59%, 0.01%, and 0.83%, respectively, verifying the
high reliability of the proposed method in modal identification and providing a
new technical approach for correcting spacecraft structural models.