ILL-Posed Study On Identification Of Moving Loads On Bridges

Identification of dynamic axle loads on bridges is an inverse problem of structural mechanics, and a notable feature of which is the presence of ill-posed problem. Most moving loads identification methods eventually transformed to the solution of a linear equation, and the different ways solving the equations often cause the different identification accuracy and robust noise immunity for ill-posed problem. Based on the basic theories and characteristics of moving loads identification, such as the interpretive methodâ… , interpretive methodâ…¡, time domain method and frequency-time domain method, the Preconditioned Conjugate Gradient Method (PCGM) is introduced to the force identification in this thesis. Some illustrated simulations show that the identified results of the PCGM have many advantages, such as higher identification accuracy, more robust noise immunity, faster convergence and easier selection of initial parameters and so on. Meanwhile, the solution of the ill-posed problems existed in the inverse problem of moving force identification can be improved by properly selection of preconditioned matrix in the PCGM.Some numerical simulations have been conducted, and the illustrated results show that the PCGM has higher identification accuracy and less sensitivity to noise when compared with the TDM, especially by using bending moment responses for identifying the time-varying loads alone, the identification accuracy is obviously improved than that by TDM, which is beneficial to the application of the force identification techniques in practice. Because the bending moment responses contain more low frequency components at most and have less sensitivity to the high frequency noise, therefore, the PCGM is more accurate when the bending moment responses are used to identify the moving loads, especially when the responses have been polluted by noise. Otherwise, the preconditioned matrix, which affects identification accuracy and robust noise immunity as well as robust ill-conditioning problem to some extent, is very important to the PCGM.In order to evaluate the PCGM and apply it to application, a series of vehicle-bridge models were fabricated in laboratory, some experiments were further conducted. The bending moment responses were measured under different cases, the moving vehicle loads on bridges are identified by using the PCGM. Some parameter studies show that the identified results based on the PCGM have higher identification accuracy, faster convergence and more robust immunity to the ill-conditioning problems on the moving force identification. The results indicate that the PCGM can identify moving force in most cases with high accuracy, it means that the PCGM is valid to identify moving loads on bridge deck and can be used in fields. At the same time, the parameters of bridge and vehicle have a significant influence on identification accuracy, choosing proper cases can improve the accuracy obviously. Finally, some conclusions were made, recommendations proposed for further studies in the future.