关键词: 知识-数据驱动, 沉管隧道, 管节接头, 安全状态, 仿真分析, 神经网络

Abstract: The safety status of immersed tube tunnel(ITT) joints is continuously evolving, making direct deformation sensing challenging. Therefore, a case study is conducted on the subsea ITT of the Hong Kong-Zhuhai-Macao bridge. By examining existing monitoring data, the deformation patterns of ITT joints are summarized to reveal the strong correlation between the expansion/contraction deformation of joints and structural temperature as well as the significant impact of tidal variations on the shear deformation at joints. Furthermore, a knowledge-data-driven rapid inference method is proposed for joint deformation. This method involves constructing a refined finiteelement model of the tunnel, analyzing its mechanical behavior under various typical deformation modes, and generating a dataset of deformation service behaviors. By employing a back-propagation neural network, a comprehensive cross-sectional deformation inference model is developed for tunnel joints based on the simulated joint service behaviors. This enables the rapid reconstruction of joint deformations using limited actual measurement data. The proposed analytical method has been successfully applied to the on-site maintenance of the ITT of the Hong Kong-Zhuhai-Macao bridge. The case study demonstrates that during Typhoon Saola in 2023, the overall impact of the typhoon on the ITT joint system was relatively small.

Key words: knowledge-data-driven, immersed tube tunnel, pipe joint, safety status, simulation analysis, neural network