Study On Structural Modal Parameter Identification And Its Application In The Shaking Table Model Test

Although modal parameter identification is not the final goal for many engineering issues, it is the basis and prerequisite for some current research frontiers, such as structural dynamic modification, structural health monitoring, damage detection. Therefore, it has been paid considerable attention in mechanical engineering, aerospace engineering, civil engineering among others. Although a lot of identification algorithms and corresponding modified forms have been developed, there are still some problems in terms of their accuracy and effectiveness. For example, for rational fractional polynomial method and orthogonal polynomial method, which are based on fitting measured frequency response functions, the former is disturbed by ill solution and misfit while the latter is obsessed by precision loss with transformation of base function. On the other hand, for algorithms in time domain, such as eigensystem realization algorithm and stochastic subspace identification, construction of Hankel matrix and determination of system order are the main issues with which no general methods to deal up to now.In order to compare the similarities and discrepancies of different modal identification algorithms, the typical identification algorithms are reviewed and some of them are focused and programmed using Matlab. Furthermore, the numerical simulation results of a 3-storey shear-type reinforced concrete frame and the data taken from a shaking table test on a 12-storey reinforced concret frame model are processed using these algorithms and the corresponding programs. The comparison of the identification results is further discussed.(1) The development of modal parameter identification technique and the basic principles and the application potential of typical identification methods are reviewed. In order to facilitate the algorithm selection, the similarities, discrepancies and applicable condition of various algorithms are summarized for both frequency domain methods and time domain methods. The former includes rational fraction polynomial and pick peaking technique and the latter includes system realization algorithm and complex exponential method.(2) Taken rational fraction polynomial method in frequency domain and eigensystem realization algorithm in time domain as the identification methods, the problems of two algorithms are discussed and the corresponding solutions are proposed. For the lower order system, the problem of order missing for RFP method is solved by using amplitude normalized weight function or imaginary-part normalized weight function. For ERA, the problem with respect to constructing Hankel matrix and determining the system order is solved by building square matrix and 3 methods for determing order. Identification results for a numerical simulation example and an experiment example indicate that the proposed methods and programs are reliable.(3) Taken the data from a shaking table test on a 12-storey reinforced concrete frame model by State Key Laboratory for Disaster Reduction in Civil Engineering, Tongji University as an example, the proposed methods and programs are used to identify the model modal parameters. This model is also analyzed by finite element software. The identification results and the comparison with finite element analysis results further consolidate the reliability and effectiveness of the proposed methods and the corresponding programs.