Theory And Experiment On Moving Load Identification Of Complex Bridge Structures

Fatigue damage and failure may be easily occurred in bridge structures under the repeated action of moving vehicle loads and the effect of traffic overloads, which may influence the normal service life severely. The real-time detection and record on ambient loads environment are also required in the health monitoring system of bridges. Therefore, to identify the moving vehicle loads on bridges has important theoretical significance and engineering value for the health monitoring and daily maintenance of bridges and the traffic planning.A systematic investigation on identification of moving loads on complex bridges such as multi-span continuous girder, spatial girder and curved box girder is performed in this dissertation. The main researches and innovations are included as follows:(1) Based on the modal superposition method and the exact solution of natural vibration of a continuous beam, the equations of motion of a uniform continuous girder subjected to a set of moving loads are established. Dynamic responses of the bridge are approached using the spline function approximation method, and the regularized solutions of the moving loads are obtained by combining the Tikhonov regularization with the singular values decomposition (SVD).(2) Against bridges with complex structural layouts, a finite element model of the bridge structure is established by employing the quintic Hermitian interpolating function as an element shape function. In the modal space, the Chebyshev orthogonal polynomial is taken as the time finite element shape function of responses and forces, and the time finite element model for moving load identification is established based on the weighted-residual method. Then the stable solutions of moving loads are obtained using the regularization method with truncated singular values decomposition.(3) Against the spatial girder bridges in general use in highway bridges and urban overpasses, a spatial grillage model for moving load identification of bridges is established. By employing the discrete vibration modes of the bridge, the stable solutions for moving load identification is obtained by means of the spline function approximation method and the truncated singular value decomposition regularization method, in which the measuring point placements are optimized using the successive accumulation method in which the modal assurance criterion (MAC) matrix is taken as a target function.(4) Against curved box girder bridges, a grillage model of the curved box girder is built based on the shear-flexible grillage method, and the vehicle-bridge interaction is investigated. A novel curved beam element considering shearing deformation and rotational inertia is derived using the trigonometric function and based on the equilibrium equation. A spatial vehicle model is established according to the Lagrange principle, and four grades of road irregularity are generated based on the inverse Fourier transforming technique.(5) A novel BP neural network-based identification method for moving loads of bridges is proposed. Using the stage identification technique, the positions, velocities, distances and loads of vehicles is identified step by step. The design of samples set, training algorithm, and the initial values of weights of the neural network are improved and optimized by using the orthogonal design method, the regularization method and the genetic algorithm respectively.(6) A simply supported steel beam with 2 m length and a model car are fabricated in laboratory, considering the similarity conditions. Bending stiffness of model beam, measurement of strain and speed of car are calibrated. The first four modes and its modal parameters of the model beam are obtained by test. Through the measured strains and modal parameters, the proposed methods for moving load identification of bridges are validated and parameter investigated.