Study On Structural Damage Identification Method Based On Hierarchical Bayesian Model Updating

Due to the influence of factors such as observational noise,modeling errors and service environment,structural models and observations both have strong inherent uncertainties.The model updating method based on hierarchical Bayesian theory can take the structural parameters themselves as random variables through introducing the over parameters to consider the intrinsic uncertainty of structural parameters caused by environmental factors,and it has potential to become a kind of uncertain structural damage identification method superior to the traditional Bayesian method.This paper will aim at selecting the parameters that are more sensitive to structural damage or have more extensive application scope as the optimization target of the correction algorithm,thus improving the hierarchical Bayesian model updating method to construct a hierarchical Bayesian model correction framework with higher recognition accuracy and efficiency,and developing the corresponding structural damage identification methods.The main contents are as follows:(1)A hierarchical Bayesian structural model updating framework for identifying structural physical parameters is constructed based on hierarchical Bayesian theory.Taking a three-layer shearing frame as an example,the effectiveness and identification accuracy of structural damages respectively with traditional modal parameters and time-domain dynamic responses as hierarchical Bayesian model-updating targets are compared,and then the recognition accuracy and anti-noise performance of the hierarchical Bayesian model updating algorithm are compared with that of the traditional Bayesian model correction algorithm by using structural dynamic responses as updating target.(2)To overcome the problem of low computational efficiency caused by large redundancy of structural dynamic response information,a hierarchical Bayesian structural damage identification algorithm is proposed with the feature parameters extracted from structural dynamic responses as updating targets.Taking a three-layer shearing frame as a numerical example,through damage sensitivity analysis and comparison of damage identification accuracy and efficiency,the 2-norm of structural dynamic response is selected as ideal feature parameters,and then the effectiveness of the improved hierarchical Bayesian algorithm is verified by using the numerical example and the LANL eight-degree-of-freedom mass spring Benchmark model test.(3)To overcome the problem that actual random natural excitation information is difficult to be completely tested,a hierarchical Bayesian structural damage identification method based on structural dynamic response correlation function is proposed.By preprocessing structural dynamic response and its related functions,a smooth free vibration response of structures is obtained as the updating target.The effectiveness and applicability of the algorithm are studied by using two numerical examples including a three-layer shearing frame and the LANL mass spring Benchmark model.