| Recently, the health monitoring of structures has been an important research field in civil engineering development. Meanwhile, model updating methodology based on system identification procedures has become a hot topic in structural damage detection.In this paper, the substructure approaches of structural model updating are studied. The main points are such as follows:First of all, the number of updated elements'DOFs(degree of freedom) are reduced on the direction of measured signals, and the correlations among the reduced elements are analyzed, through which the correlations of the influences on the measured mode are estimated. The principle of dividing substructure is established by dividing the elements, which have high correlation of the influences, into one substructure.Secondly, the elements of mass orthogonal matrix and stiffness orthogonal matrix are multiplied by a matrix with zero on diagonal line and one rest. Then the objective functions are established for optimal solution. In this way, the substructure method of structural model updating is improved, which breaks through the limitation of demanding normalized amplitudes with regard to mass.Thirdly, take population obtained from genetic algorithm as the initial values, the updating factors can be got by substructure method. This method can solve the problem that the solutions tend to trap in local optimum when the number of updating factors is more and the measured modes are less.Fourthly, the conversion relationship between strain mode and displacement mode are established in terms of beam elements, and strain mode is applied on updating model to expand the application domains of substructure updating method from displacement mode space to strain mode space. Then displacement mode and strain mode are incorporated rationally to update structural model.Finally, a simulation studies was carried out on the frame structure presented in the paper, which indicated the feasibility and availability of the method. |
PDF Full Text Request