Theoretical And Experimental Study Of Random Loading Identification Of Structures

The technology of structural dynamic loading identification is one of the key problems in modern engineering designs, and has a wide application future. Loading identification, like the system parameter identification and dynamic characteristics modification, is an inverse problem of structural vibration. Although the classical theory of dynamic characteristics and response analysis for a vibration system has been well established and is playing an important role in engineering applications, the research on loading identification still has a far way to go. The inverse problem of vibration, differing from the analysis method of a classical direct problem, is a combined method of experimental and theoretical analyses to deal with the vibration problems in engineering, i.e., to obtain the dynamic loading exerted on the system from the dynamic characteristics and measured dynamic responses. Comparatively, the experimental means play a more important role in this research field.In this thesis the efficient and accurate algorithm to obtain the random response power spectrum, the so called Pseudo Excitation Method (PEM), is inversely extended, so that some ordinary methods can be used in the solution of the problem of random loading spectrum identification for proportionally or non-proportionally damped structures. In other words the method can predict the auto-spectral density and cross-spectral density of the excitations if the auto-spectral density and cross-spectral density of a few responses of the system are measured in advance. Such responses can be either displacements, velocities, accelerations, strains and so on, or a mixed combination of them. This gives rise to many chooses for the types and locations of the measured responses. In order to conduct the measurements and identifications more effectively, the effects of the measuring locations, measuring point number, structure damping, and measuring accuracy, etc, on the loading spectrum identification are studied by means of numerical simulations. The study shows that computer simulation used herein to select optimized measuring responses is one of the very efficient methods to eliminate or reduce the ill-condition in the frequency response matrix.The ill-condition of the frequency response matrix has long been one of the key problems in loading identification. Usual methods to cope with this difficulty depend either on the use of very expensive high precision instruments, or on the development of special computation technologies for the solution of ill-conditioned equations. It has been more and more difficult to improve the effect in terms of these means. A technique to use temperate and additional damping is proposed in the present paper to reduce the ill-condition of the frequency response function, which appears to be a good alternative for such problems.So far, theoretical or experimental studies of random loading identification have rarely been seen in the literature. Based on the theoretically established Inverse Pseudo Excitation Method (IPEM), a systematic experiment is also designed and accomplished to test the correctness and effectiveness of IPEM. The comparison between the experimental results and the computer simulations show that the IPEM is a reliable, simple and efficient method for the identification of stationary random loading spectrum acted on a linear time-invariant system. The identification precision is high enough to meet general engineering requirements. It is found that the method has a good development future.The precise integration method is based on the innovative precise solution strategy for the computation of exponential matrices. It gives an unconditionally stable explicit integral format. Its time step size is not restricted by the eigenvalues of the structure under consideration. A high precision solution can be obtained so long as the load inside the integration step length changes in a linear, harmonic, or many other deterministic forms. This paper also inversely extends the precise integration method by proposing a loa...