Finite Element Model Update via Bayesian Estimation and Minimization of Dynamic Residuals

An algorirlvn is presentedfor updatingfinite element mod & based upon a minimization of dynamic residuals. The dy-namic residual of interest is theforce unbalance in rhe homogeneous form of r/z equations of motion arising from errors in rhe model’s moss and srifiess when evaluated with the identified modal poram-efers. The present olgorirhm is n modification ami extension of o previously-developed Sensitivily-Bared Element-By-Element (SB-EBEI merhod for damage daecrion nndfinile element model updar-ing. In lhe present olgoritbm. SB-EBE has been generalized to min-imize a dynamic dispplncement residual quamily, which is shown lo improve test-analysis mode correspondence. Furrhumore. the algo-rithm has been modified to include Bayesion estimation conceprs, and rhe underlying nonlinear optimizarionproblem has been consis-rently linearized IO improve rhe convergence properries. The resulr-img algorithm is demofzstrated via numerical arpd experimental examples to be an efficient and robust m&ad for both localizing model errors and esrimating physical parameters.