In order to improve the calculated accuracy of finite element model (FEM) for a large-scale aqueduct damage diagnosis, a FEM updating method based on information fusion and response surface method (RSM) is proposed. The initial FEM method from the aqueduct of one project in china is updated. The parameters of the design space and the frequencies by central composite design (CCD) are built up. Moreover, the dynamic response of the aqueduct under environmental excitation is measured. The test signals that cover three spans of aqueduct are processed to obtain complete and effective frequencies by a signal denoising method based on complete ensemble empirical mode decomposition with adaptive noise-singular value decompose(CEEMDAN-SVD) and an information fusion algorithm based on variance contribution rate. The first four modal frequencies are extracted as the target to optimize the response surface, and the modal frequency of the updated FEM method is compared with the frequency after processing based on structure vibration test. The results show that the calculated frequency of the updated FEM method and the measured frequency are in a good agreement. This study is a successful application of the FEM updating technique on a large-scale aqueduct structure, and provides a baseline evaluation model for the structural damage diagnosis and health monitoring of aqueduct structure.