Bridge influence line identification using an adaptive enhanced variational mode decomposition

To reduce the errors resulting from the dynamic effect and noise, a novel method is proposed for bridge influence line (BIL) identification. Firstly, the classical variational mode decomposition is developed into an adaptive enhanced variational mode decomposition (AEVMD) by integrating odd extension and modal confidence criteria. Then, the AEVMD method is employed to extract the quasi-static responses of the bridge under moving vehicular loads. Finally, the preprocessed conjugate gradient least squares is applied to solve the inverse problem of BIL identification. The proposed method is validated through numerical analysis, investigating the effects of road roughness, measurement noise, vehicle type, and vehicular speed. The results demonstrate that the AEVMD can accurately extract the quasi-static bridge responses in an adaptive manner while overcoming the end-point effect. The proposed BIL identification method can achieve higher accuracy and stability, and the mean error of displacement and strain influence lines is less than 2% and 6%, respectively. Moreover, the effectiveness and superiority of the proposed method are verified through both laboratory and field tests. The BIL errors for laboratory tests are within 3%, and the BIL is reasonably identified even at vehicular speeds exceeding 80 km/h in the field tests. These results further demonstrate the robustness of the proposed approach.