Bridge early warning based on structural health monitoring (SHM) system is of significant importance for ensuring bridge safe operation. The temperature-induced deflection (TID) is a sensitive indicator for performance degradation of continuous rigid frame bridges, but the time-lag effect makes it challenging to predict the TID accurately. A bridge early warning method based on nonlinear modeling for the TID is proposed in this article. Firstly, the SHM data of temperature and deflection of a continuous rigid frame bridge are analyzed to examine the temperature gradient variation patterns. Kernel principal component analysis (KPCA) is used to extract principal temperature components. Then, the TID is extracted through wavelet transform, and a nonlinear modeling method for the TID considering the temperature gradient is proposed using the support vector machine (SVM). Finally, the prediction errors of the KPCA-SVM algorithm are analyzed, and the early warning thresholds are determined based on the statistical patterns of the errors. The results show that the KPCA-SVM algorithm achieves high-precision nonlinear modeling for the TID while significantly reducing the computational load. The prediction results have coefficients of determination above 0.98 and fluctuate within a small range with clear statistical patterns. Setting the early warning thresholds based on the statistical patterns of errors enables dynamic and multi-level warnings for bridge structures.

This paper introduces an innovative model for heavy-haul train–track–bridge interaction, utilizing a coupling matrix representation based on the virtual work principle. This model establishes the relationship between the wheel–rail contact surface and the bridge–rail interface concerning internal forces and geometric constraints. In this coupled system’s motion equation, the degrees of freedom (DOFs) of the wheelsets in a heavy-haul train lacking primary suspension are interdependent. Additionally, the vertical and nodding DOFs of the bogie frame are linked with the rail element. A practical application, a Yellow River Bridge with a heavy-haul railway line, is used to examine the accuracy of the proposed model with regard to discrepancy between the simulated and measured displacement ranging from 1% to 11%. A comprehensive parametric analysis is conducted, exploring the impacts of track irregularities of varying wavelengths, axle load lifting, and the degradation of bridge stiffness and damping on the dynamic responses of the coupled system. The results reveal that the bridge’s dynamic responses are particularly sensitive to track irregularities within the wavelength range of 1 to 20 m, especially those within 1 to 10 m. The vertical displacement of the bridge demonstrates a nearly linear increase with heavier axle loads of the heavy-haul trains and the reduction in bridge stiffness. However, there is no significant rise in vertical acceleration under these conditions.

This study proposed a separation method to identify the temperature-induced response from the long-term monitoring data with noise and other action-induced effects. In the proposed method, the original measured data are transformed using the local outlier factor (LOF), and the threshold of the LOF is determined by minimizing the variance of the modified data. The Savitzky–Golay convolution smoothing is also utilized to filter the noise of the modified data. Furthermore, this study proposes an optimization algorithm, namely the AOHHO, which hybridizes the Aquila Optimizer (AO) and the Harris Hawks Optimization (HHO) to identify the optimal value of the threshold of the LOF. The AOHHO employs the exploration ability of the AO and the exploitation ability of the HHO. Four benchmark functions illustrate that the proposed AOHHO owns a stronger search ability than the other four metaheuristic algorithms. A numerical example and in situ measured data are utilized to evaluate the performances of the proposed separation method. The results show that the separation accuracy of the proposed method is better than the wavelet-based method and is based on machine learning methods in different time windows. The maximum separation errors of the two methods are about 2.2 times and 5.1 times that of the proposed method, respectively.

During the operation of fabricated small box girder bridges, which face safety issues such as structural degradation and failure, there is an urgent need to propose a safety evaluation method to cope with the possible risks. This article quantitatively evaluates the safety state of a fabricated small box girder bridge in Wuhan City based on Fuzzy Analytic Hierarchy Process (FAHP) and structural health monitoring (SHM) data. Firstly, the FAHP model is established, and stress, deformation, and temperature are selected as evaluation factors. The safety thresholds of stress and deformation are determined by combining the industry specifications and the historical statistical patterns of the massive SHM data. The temperature field of the bridge is simulated and analyzed by combining ANSYS, HYPERMESH, and TAITHREM, and the most unfavorable temperature gradient is determined as a threshold for the safety evaluation. Finally, the scores of indexes of the bridge are determined based on the measured SHM data, which in turn provides a quantitative description of the safety state. The results show that the thresholds determined by the joint industry specifications and the massive SHM data are reasonable; the temperature field simulation model established in this article is consistent with the measured results, and can accurately determine the temperature gradient of the bridge. The safety evaluation result from the FAHP model is the same as the field test results, which verifies the effectiveness and applicability of the proposed method to actual bridge projects.

During deep excavation, actual soil parameters undergo changes. To enhance the accuracy of soil parameter selection in finite element simulation and improve the precision of finite element analysis, an inversion method for soil parameters based on a PSO-SVM model is proposed. In this method, the particle swarm optimization (PSO) algorithm is utilized to optimize the penalty parameter C and kernel function parameter g of the support vector machine (SVM) model. The optimized PSO-SVM model is employed to establish a nonlinear mapping relationship between the horizontal displacements of retaining structures in deep excavations and soil parameters through orthogonal experimental design and finite element simulation analysis. Subsequently, soil parameters are inverted from monitoring data of horizontal displacements of retaining structures, and the reliability of the parameters is verified. The deformation of the retaining structures during subsequent cases is then predicted. The results demonstrate that the absolute error of the peak maximum horizontal displacements of the retaining structures after inversion is maintained within 1 mm. The maximum relative error is reduced from 18.96% before inversion to 7.63%, indicating that the inverted soil parameters for the deep excavation possess high accuracy. The precision of the finite element simulation for deep excavation is significantly improved, effectively reflecting the actual mechanical properties of the soil during the construction stage. The inverted parameters can be used for the prediction of subsequent retaining structure deformation. During subsequent construction conditions, the predicted maximum horizontal displacement (deformation) of the retaining structure at monitoring point CX1 is 15.66 mm, and that at monitoring point CX2 is predicted to be 14.22 mm. Neither value exceeds the project warning threshold of 30.00 mm.

Accurate finite element (FE) models are essential for the safety assessment of civil engineering structures. However, obtaining reliable model parameters for existing bridges remains challenging due to the inability to conduct static load tests without disrupting traffic flow. To address this, this study proposes an FE model updating framework that integrates the response surface method and the nutcracker optimization algorithm (NOA). This framework is characterized by the incorporation of ambient vibration data into parameter optimization, thereby enhancing model accuracy. The stochastic subspace identification method is first adopted to extract the bridge’s natural frequencies from vibration data. The response surface method is then employed to construct a response surface function that approximates the FE model. The NOA is subsequently applied to iteratively optimize this response surface function, ensuring rapid convergence and the precise adjustment of the FE model parameter. To validate the effectiveness of the proposed framework, a continuous beam–arch composite bridge with a span of 204.783 m was selected as a case study. The results indicate that the proposed method reduced the average frequency error from 5.58% to 2.75% by updating the model parameters. While the whale optimization algorithm required 21 iterations and the grey wolf optimizer needed 41 iterations to converge near the minimum, the NOA achieved this in merely 13 iterations, demonstrating the NOA’s superior convergence speed. Furthermore, the NOA significantly outperformed both the whale optimization algorithm and the grey wolf optimizer in reducing the error of the first transverse vibration frequency.