Identification of Structural Systems with Full Characteristic Matrices under Single Point Excitation M. Ghafory-Ashtiany, B. Adhami & K. Khanlari Department of Civil Engineering, Science and Research Branch, Islamic Azad University (SRBIAU), Tehran, Iran SUMMARY The aim of "System Identification" is to determine the modal and system properties of structural systems. Because of various constraints in practice only single excitation and partial measurement at selected degrees of freedom is possible. In this paper, to identify a structural system, dynamic load was applied only along one of the degrees of freedom of the structure and the responses corresponding to a few degrees of freedom have been measured. To identify characteristic matrices of a system with this sort of restricted information, a new approach was intrtoduced. Taking into account the significant effect of noise reduction in improving the system identification accuracy levels, a noise reduction technique was also proposed. It was shown that as noise level increases, identification errors will also increase though to an acceptable range. The method's efficiency and precision were examined through the application of a "closed loop solution" to a six-storey flexural structure. Keywords: System identification; Single force; Identification vector; Tree-coefficient-matrices; Noise reduction 1. INTRODUCTION Structural Health Monitoring (SHM), as a need for reliable assessment of structural safety under service or extreme loads such as earthquake, requires system identification and damage detection. System Identification (SI) determines structural dynamic characteristics such as modal properties (frequencies, mode shapes and damping ratios) and system properties (mass, damping, and stiffness matrices). System identification methods that directly identify the characteristic matrices of a system (M, C, and K), have been investigated over several years as follows: Potter and Richardson in 1974, Richardson in 1977 as well as Richardson and Shye in 1987 developed an approach to identify mass, damping, and stiffness matrices of a linear elastic system. In the method, using Laplace Transform of the measured input, and the displacements along the system’s degrees of freedom (DOFs) under arbitrary loadings, and the “Transfer Matrix” (as a binomial function in term of Laplace value with coefficient of mass, damping, and stiffness matrices), they proposed a formulation for the direct identification of the system’s characteristic matrices. They applied full force vector on each DOF. The result for noise free case was exact and correct but their method did not include a discussion of the identification errors due to noise (Potter and Richardson, 1974, Richardson, 1977, Shye and Richardson, 1987). Masri, Miller, Saud and Caughey in 1987, and Agbabian, Masri, Miller and Caughey in 1991 studied the direct identification of characteristic matrices of linear and nonlinear structural systems, and presented a detailed formulation which was based on the inverse solution of the problem of identifying the system's characteristic matrices in time domain. The investigated structure in their method was a special case of shear structure, and hence its matrices of stiffness and damping were trigonal. In their method, the mass matrix was assumed to be known, and by this unreal assumption, the identification errors in two other matrices (damping and stiffness) largely decreased (Masri et al., 1987, Agbabian et al., 1991).