To analyze the kinetic characteristics of a debris flow that occurred on 9 April 2000 in Tibet, China, a meshfree numerical method named smoothed particle hydrodynamics (SPH) is introduced, and two-dimensional and three-dimensional models are established in this work. Based on the numerical simulation, the motion process of this debris flow is reproduced, and the kinetic characteristics are analyzed combining with the field investigation data. In the kinetic analysis, the flow velocity, runout distance, deposition, and energy features are discussed. Simulation results show that the debris flow mass undergoes an acceleration stage after failure, then the kinetic energy gradually dissipates due to the friction and collision during debris flow propagation. Finally, the debris flow mass blocks the Yigong river and forms a huge dam and an extensive barrier lake. The peak velocity is calculated to be about 100 m/s, and the runout distance is approximately 8000 m. The simulation results basically match the data measured in field, thus verifying the good performance of the presented SPH model. This approach can predict hazardous areas and estimate the hazard intensity of catastrophic debris flow.

Submarine debris flows and their generated waves are common disasters in Nature that may destroy offshore infrastructure and cause fatalities. As the propagation of submarine debris flows is complex, involving granular material sliding and wave generation, it is difficult to simulate the process using conventional numerical models. In this study, a numerical model based on the smoothed particle hydrodynamics (SPH) algorithm is proposed to simulate the propagation of submarine debris flow and predict its generated waves. This model contains the Bingham fluid model for granular material, the Newtonian fluid model for the ambient water, and a multiphase granular flow algorithm. Moreover, a boundary treatment technique is applied to consider the repulsive force from the solid boundary. Underwater rigid block slide and underwater sand flow were simulated as numerical examples to verify the proposed SPH model. The computed wave profiles were compared with the observed results recorded in references. The good agreement between the numerical results and experimental data indicates the stability and accuracy of the proposed SPH model.

Peridynamics (PD) is a novel nonlocal theory of continuum mechanics capable of describing crack formation and propagation without defining any fracture rules in advance. In this study, a multi-grid bond-based dual-horizon peridynamics (DH-PD) model is presented, which includes varying horizon sizes and can avoid spurious wave reflections. This model incorporates the volume correction, surface correction, and a technique of nonuniformity discretization to improve calculation accuracy and efficiency. Two benchmark problems are simulated to verify the reliability of the proposed model with the effect of the volume correction and surface correction on the computational accuracy confirmed. Two numerical examples, the fracture of an L-shaped concrete specimen and the mixed damage of a double-edged notched specimen, are simulated and analyzed. The simulation results are compared against experimental data, the numerical solution of a traditional PD model, and the output from a finite element model. The comparisons verify the calculation accuracy of the corrected DH-PD model and its advantages over some other models like the traditional PD model.

In order to comprehensively investigate the spatio-temporal dynamics of corrosion evolution under complex pipeline environments and improve the corrosion rate prediction accuracy, a novel framework for corrosion rate prediction based on adaptive multivariate time series graph neural network (MTGNN) multi-feature spatio-temporal correlation analysis is proposed. First, pipeline monitoring points are modeled as graph nodes to construct the pipeline corrosion spatio-temporal information graph, with corrosion rate and auxiliary features (selected through feature correlation analysis) forming node attributes. Then, a dynamic adjacency matrix is adaptively learned to capture hidden spatial dependencies, while temporal convolution modules extract multi-scale temporal patterns, and the node sequences with integrated corrosion features are input into the adaptive MTGNN for prediction. To reduce the accumulation of errors in multi-step prediction, a “chunked progressive” training strategy is adopted, incrementally expanding prediction horizons. Finally, experiments based on real urban drainage pipeline data show that in six-step predictions, the model reduces MAE, RMSE, and MAPE by 6.59–32.16%, 4.38–27.95%, and 5.01–22.22%, respectively, compared to traditional time series methods such as Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and non-adaptive MTGNN. The results indicate that the adaptive MTGNN, which integrates multi-source node features, has higher prediction accuracy across the three evaluation metrics, highlighting its capability to leverage spatio-temporal synergies for accurate short-term corrosion rate prediction.

Accurately predicting stress in large-diameter sewage pipelines is critical for ensuring their structural reliability and safety. To meet the safety requirements of large-diameter concrete pipes, we propose a hybrid model that integrates Kolmogorov-Arnold Networks (KAN) with Long Short-Term Memory (LSTM) neural networks. The model is trained and validated using actual pipeline monitoring data, ensuring that it accurately captures both the temporal dependencies and nonlinear stress patterns inherent in such systems. By modifying the fully connected layer of the original LSTM model, we develop a novel LSTM-KAN model and evaluate its performance through comprehensive predictive analysis. Comparisons with a traditional LSTM model reveal that the LSTM-KAN model—in which the fully connected layer is replaced by a KAN layer—achieves significantly lower loss and higher accuracy with fewer training iterations. Specifically, the proposed model attains a mean absolute error (MAE) of 0.033, a root mean square error (RMSE) of 0.035, and a coefficient of determination (R²) of 0.92, underscoring its superior accuracy and efficiency, and it can be used for the long-term prediction of stress in large-diameter pipes. Moreover, the integration of KAN significantly improves the nonlinear modeling capacity of the conventional LSTM, enabling the hybrid model to effectively capture complex stress variations under variable operating conditions. This work not only provides novel technical support for the application of deep learning in pipeline stress prediction but also offers a robust framework adaptable to other structural health monitoring applications.