Tunnel settlement deformation monitoring is a complex task and can result in nonlinear dynamic changes. To overcome the disturbances caused by historical data and the difficulty in selecting input parameters during deformation prediction, a decomposition, reconstruction and optimization method for tunnel settlement deformation prediction is proposed. First, empirical mode decomposition (EMD) is used to decompose the in-situ monitoring data and reduce the interactions among information at different scales in sequences. Then, the monitoring data are decomposed into intrinsic mode functions (IMFs). Secondly, the smoothing factor of the generalized regression neural network (GRNN) is optimized by using the sparse search algorithm (SSA). An EMD-SSA-GRNN deformation prediction model is developed using the optimized GRNN algorithm and is used to predict the changes in the decomposed IMFs. Finally, using the measured deformation data from a shallowly buried tunnel along the Kaizhou-Yunyang Highway in Chongqing, China, the reliability and accuracy of different models are analysed. The results show that tunnel settlement deformation exhibited a trend and a slow change in the early stage, a rapid change in the middle stage and a slow change in the late stage, and the rate of change was significantly influenced by the excavation time and the upper and lower geological layers. The prediction accuracy of the EMD-SSA-GRNN model after EMD improved from 19.2 to 59.4% relative to that of the SSA-GRNN and single GRNN models. Moreover, we find that the three error evaluation indicators of the EMD-SSA-GRNN model are lower than those of the other models and that the results of the proposed model and are more strongly correlated with measured data.

Based on a prototype of the Beijing subway tunnel, this research conducts large-scale model experiments to systematically investigate the vibration response patterns of tunnels with different damage levels under the influence of measured train loads. Initially, the polynomial fitting modal identification method (Levy) and the model test preparation process are introduced. Then, using time-domain peak acceleration, frequency response function, frequency-domain modal frequency, and modal shape indicators, a detailed analysis of the tunnel’s dynamic response is conducted. The results indicate that damage significantly amplifies vibration acceleration, with the amplification increasing with the severity of the damage. When the crack lengths are 2 cm, 4 cm, and 6 cm, the peak acceleration increases by 25.12%, 36.35%, and 50.29%, respectively, while adjacent segments show increases of 13%, 29%, and 45%. Damage decreases the tunnel structure’s modal frequency, with the first two modal frequencies showing the most significant reductions of 9.87% and 7.34%, respectively. The adjacent segments show reductions of 7.7% and 4.2%. As the severity of the damage increases, the amplitude of the modal shape at the damaged location also increases, with the first modal shape rising by 43.37% for 4 cm damage compared to 2 cm damage and by 72.21% for 6 cm damage. The second modal shape increases by 9.04% and 26.51%, respectively. Additionally, the effectiveness of the polynomial fitting modal identification method (Levy) for tunnel structural damage detection was validated. Finally, based on the methods outlined above, the tunnel responses measured on-site in the Beijing metro were also analyzed. The findings of this study provide important theoretical support for the assessment and routine maintenance of metro tunnels.

Expansive soil is characterized by its unique structural morphology and drastic volume change. With infrastructure increasingly constructed in expansive soil areas, engineering problems caused by the properties of expansive soils have attracted more attention. Cyclic wetting‐drying and shear testing were accordingly conducted on an expansive soil from Chengdu area in China. Crack development and shear strength change were analyzed using the Mohr–Coulomb equation for shear strength by fitting the experimental data. The results show the following: (1) With the increase in wetting‐drying cycles, the crack ratio increases, the shear strength decreases, and the shear strength parameters gradually decrease at the same rate of change. The applied vertical load reduces the weakening effect of the wetting‐drying cycles on the soil structure and strength by restraining the expansion and contraction deformation. (2) By analyzing the number of wetting‐drying cycles and the crack images, the crack development (length, direction, etc.) of the expansive soil can be predicted and described. (3) There is a specific linear correlation between the crack ratio and strength that approached a limit value with ongoing wetting‐drying cycles. The strength of the expansive soil can therefore be obtained based on crack development, improving the ability of designers to account for the behaviour of expansive soils.

This paper proposes a novel parameterized frequency‐domain modal parameter identification method, called direct modal variational mode decomposition (DMVMD), based on the multivariate variational mode decomposition (MVMD) framework and the principle of modal superposition. Under the constraint of normalized mode shapes, this paper theoretically derives the relationship between multivariate variational mode decomposition and the natural frequencies and mode shapes of structural systems. The aim is to extract K response modes and their corresponding mode shapes from the excited C‐dimensional vibration signals of the measured component’s response. First, the measured multichannel vibration signals are decomposed into IMFs aligned with K‐order natural frequencies using multivariate variational mode decomposition (MVMD). Then, the Hilbert equations and mode shape normalization constraints are used to solve the structural natural frequencies and mode shapes. Furthermore, the proposed multimodal identification algorithm has been validated through numerical simulations and experimental examples, demonstrating its high accuracy and robustness in modal identification. Compared to the existing multimodal algorithms related to variational mode decomposition, the proposed method is more direct and elegant. This method has been successfully applied to the modal parameter identification of subway tunnel structures, enabling accurate determination of the location of tunnel damage through analysis of the identified modal parameters.