An investigation of system identification with eccentric mass shakers

Abstract

This thesis deals with the problem of system identification of a structure tested with eccentric mass shakers. The analysis methods studied for this purpose are Eigensystem Realization Algorithm applied with Auto-Regressive Exogeneous models, Frequency Domain Decomposition technique and Covariance Driven Stochastic Subspace method. The Eigensystem Realization Algorithm is analyzed for the scenerio when the input and the output data can be measured in eccentric mass shaker expreiments. The Frequency Domain Decomposition is used for the experiments in which it is not possible to measure the phase angle between the input and the output. The Covariance Driven Stochastic Subspace method is another output-only identification scheme analyzed for making a sinesweep experiment with eccentric mass shakers. First, modal parameter identification capabilities of three methods are illustrated by a Monte Carlo analysis is handled with the data simulated from a spring-dashpot model. Then a finite element analysis program to simulate a forced vibration test conducted with eccentric mass shakers. To apply the proposed methods, necessary modifications in testing an actual building are introduced. Finally, the results of the simulation of the experiments are introduced.