Model Parameter Identification Of Bridge Based On Local Mean Decomposition

Modal parameter identification of bridge is an important component in bridge structural health monitoring system, thus it is of significance to acquire precise structural modal parameters on bridge structural health monitoring system. The existing method of modal parameter identification is based on the processing and analysis of the vibration response data, and vibration signals which are mostly non-stationary signal often have complex composition in actual bridge A better signal processing method is needed to obtain more accurate vibration information of the modal parameters of bridge from non-stationary signal. Accordingly, a self-adaptive time frequency analysis method-Local Mean Decomposition is approached to analyze and deal with non-stationary signal, last local mean decomposition is combined with random decrement technique to identify the bridge model parameters.The main research contents are as follows:1 The basic principle and realization process of Local Mean Decomposition(LMD) are formulated, and the comparison and analysis of Empirical Mode Decomposition(EMD) and Local Mean Decomposition shows the advantages and disadvantages of LMD method, then improvements are made to these deficiencies; According to the inherent law of the signal, a self-adaptive extreme point matching extending method is proposed to improve the end effect in LMD; Basing on the statistical properties of white noise, the complementary white noise method is proposed to avoid the mode mixing problem; In order to improve the computational efficiency of LMD algorithm, Hermite interpolation function based on its superiority is applied to LMD.2 The basic principle and realization process of modal parameter identification based on LMD are described in detail, and this method is applied to the identification of simulation examples. By comparison of the identification result of a three degrees of freedom spring system and its theoretical analytic solution, the identification method is proved to be correct. A model based on finite element is established for this identification method whose applicability in bridge is verified by comparison of the identification results based on EMD and LMD.3 The identification results of stochastic subspace identification and the identification results based on LMD are compared through identifying the modal parameters of a cable-stayed experimental model bridge and a long span cable-stayed bridge, and modal parameter identification based on LMD is proved to be effective and practical in actual bridge structure.