Modal Parameter Identification Based On Frequency Domain Subspace Methods

System identification based on subspace methods has advantages such as few parametersdetermination, stable computation, simple identification process and easy programming etc.Frequency domain modal identification algorithms are investigated in this thesis on the basis ofsubspace methods. The main contents are as the following:1. Frequency domain subspace modal identification theory for both input and output data, i.e. theexperimental modal analysis (EMA) technique based on subspace methods, is studied in detail. Acomparison analysis is made between the identification algorithms based on input-output data and theone based on frequency response function;2. Frequency domain subspace modal identification theory based on output-only data, i.e. theoperational modal analysis (OMA) technique based on frequency domain subspace methods, isstudied in detail. A comparison analysis is made between the identification algorithms based on outputdata and the one based on positive power spectrum density;3. The singular value chart and complex mode indicator function are applied to determine theorder of system so as to simplify the identification process and improve the identification precision;4. The simulation example of a seven degree of freedom mass-spring system and a practicalexample of an aluminum plate are employed to validate the different experimental modal analysismethods based on frequency domain subspace theory; the simulation example of a three-dimensionspatial structure and the engineering measurement data of Chaobai river bridge in Beijing arepresented to validate different operational modal analysis methods based on frequency domainsubspace identification theory. Identification precision of each algorithm above is analyzed on thecases of different noise perturbation and different parameter setting.