Model Parameter Identification And Seismic Response Analysis For Small-scaled Structures

Accurately identifying structural model parameters and the establishing reliable numerical models for simulating structural responses under external loads are crucial aspects within the realm of structural health monitoring and earthquake engineering.Nevertheless,establishing reliable models for real-world structures poses significant challenges.In this dissertation,we focus on the model parameter identification and updating of small-scale steel frame models and investigate the seismic response prediction performance of the numerical models established using experimental measurements under the laboratory noise level.The main achievements are listed as following:(1)Shake table tests for small-scale structure models.In this dissertation,shake table tests were conducted for two small-scale steel frame structures,one with its base fixed and one isolated.The upper structure of the base-isolated structure is the same as the base fixed one.Both models were fixed on the shake table and tested simultaneously.The structure models were tested under different types of excitations,including white noise,sine wave,and earthquake excitations.(2)Parameter identification and update for base-fixed structure model.Modal parameters of the base-fixed structural models were first identified adopting the covariance-driven stochastic subspace identification(Cov-SSI)algorithm.However,The Cov-SSI method sometimes yielded results that did not satisfy mode orthogonality relation and eigenvalue-eigenvector relation.Consequently,a straightforward solution is to assume equal stiffness for each layer and subsequently determine the model parameters.Furthermore,to better describe the dynamic characteristics,we have proposed a two-stage modal parameter updating method.This method can ensure mode orthogonality and eigenvalue-eigenvector relation of the updated modal parameters,and can correctly embed the physical connection relations into the model.Two numerical models were then established for seismic response modeling of the base-fixed structure,one using the equalstiffness assumption,and the other with the proposed two stage model updating method.The results indicated that the model assuming equal stiffness accurately captured the fundamental frequency of the structure but led to discrepancies of higher mode frequencies,along with around 10%errors in predicting peak inter-story drift,acceleration,and shear force,as well as floor response spectra value.The updated model can also accurately simulate seismic responses of the base-fixed model.Due to accurately capturing all modal frequencies,the updated model outperforms with around 5%prediction errors in peak values inter-story drift,acceleration,and shear force,as well as floor response spectra value.(3)Parameter identification and model update for base-isolated structure models.The model parameters of the isolated structure were calibrated using the analytical solution of an equivalent single-degree-of-freedom(SDOF)system with both viscous damping and Coulomb friction under sine wave excitation.The calibrated numerical models were validated using measured input and response under both sine excitation and earthquake excitations.The results revealed that the calibrated numerical models exhibited errors within 20%for both peak values and root mean square values of isolator drifts,shear forces,when the vibration amplitude closely the calibration condition.However,prediction results for two ground motions leading to smaller vibration amplitudes showed lager errors exceeding 20%.To address this issue,the model parameters were updated using measured input and acceleration response excited by one of the two ground motions,employing the Unscented Kalman Filtering method.The updated models were adopted to simulated the response of the two ground motions,decreasing the error of inter-story displacements,shear forces,and their root mean square values.The acceleration prediction performance was also slightly improved.This dissertation also investigated the effects of different choice of damping models during the time history analysis by comparing the results obtained by numerical models adopting viscous damping and uniform damping.The comparative results indicated that different damping models may cause significant changes in the response by the time history analysis.The effect of using different damping models may differ for different structures.