A Bayesian‐Boosting Method for Probability Distribution Estimation of Structural Damping

The structural damping ratio plays a crucial role in engineering applications such as fault diagnosis and vibration monitoring. Environmental factors—particularly environmental temperature effects and excitation amplitude effects—substantially influence damping ratio identification, resulting in considerable dispersion. Consequently, the probability distribution from the classical fast Bayesian Fourier transform posterior estimation is unsuitable for engineering applications, which limits the practical use of the Bayesian Fourier transform method. To address this limitation, this paper proposes a Bayesian‐boosting method that integrates Bayesian estimation with a Boosting algorithm. The method treats the traditional Bayesian estimation as a “weak estimator,” constructs weight coefficients from the estimated density and the associated uncertainty level, and then aggregates multiple probability distributions via a Gaussian mixture model. This procedure markedly enhances the stability and accuracy of the identification process. The experimental results demonstrate that the proposed approach can accurately and efficiently approximate the probability distribution of damping ratios under different wind‐speed conditions, effectively capturing uncertainty induced by environmental variations. Moreover, it can approximate skewed and multimodal probability distributions of the damping ratio for bridge structures. Furthermore, this paper examines the relationship between vibration intensity (excitation amplitude) and the weight coefficients, thereby clarifying the physical significance of the method. Overall, the method offers a promising solution for rapidly obtaining the probability distribution of structural modal‐parameter identification using the Bayesian Fourier transform in complex environments.