Hierarchical Bayesian Estimation of System Parameters from Dynamic Responses

Abstract

Abstract.A hierarchical Bayesian approach is proposed for estimating system parameters by directly taking dynamic responses as the fixed target. The basic theories of hierarchical Bayesian model are first introduced, and then the estimation process of system parameters is illustrated. A mass-spring system of eight degrees of freedom is studied to validate the proposed method, and the effects of modeling errors, incompleteness of dynamic response data and measurement noise on the identification results are numerically investigated. The research results show that the proposed method can accurately identify physical parameters of the system with random modeling errors and measurement noise from only several dynamic responses of the system. This method can provide a new way for parameter estimation of the system with modeling errors, incompleteness of dynamic response data and serous pollution of measurement noise.