Thermal conductivity is a crucial factor for the soil, which is significantly affected by environmental conditions. Based on the variation in the thermal conductivity and the influencing factors of silty clay obtained by the freeze–thaw cycle test, this paper adopted four machine learning models optimized by particle swarm optimization (PSO), including the artificial neural network model (ANN), random forest model (RF), support vector machine model (SVM), and extreme gradient boosting model (XGBoost) to predict the thermal conductivity of the soil. Meanwhile, mean absolute error (MAE), root mean square error (RMSE), and correlation coefficient(R2) were used to evaluate the accuracy of the models. The accuracy of the machine learning model and empirical model were also compared. Then, the Monte Carlo simulation was used to analyze the stability of the models. The research results showed that the predicted performance of the machine learning models is significantly better than the empirical models. Among all the machine learning models, the R2 of the PSO-ANN model is above 0.95, while both RMSE and MAE values are below 0.02 (W·m⁻¹·K⁻¹). In addition, the stability order of the machine learning models is PSO-XGBoost, PSO-ANN, PSO-RF, and PSO-SVM. Therefore, comprehensively considering the accuracy and stability of the four machine learning models, the PSO-ANN model is recommended to predict soil’s thermal conductivity under freeze–thaw action.
Dispersive soils are widely distributed in the seasonally frozen regions of northeastern China, where hydrothermal dynamics driven by seasonal freeze–thaw (FT) cycles dominate the hydrological evolution and mechanical deterioration of soil masses, posing a serious threat to the long-term stability of hydraulic engineering in cold regions. However, the hydro–thermo–mechanical (HTM) coupled degradation mechanisms of dispersive clay from the South Nenjiang Main Canal remain poorly understood, particularly the linkage between FT-induced microstructural evolution and macroscopic mechanical behavior. In this study, low-plasticity dispersive clay specimens were subjected to 0–12 FT cycles. Unconsolidated undrained (UU) triaxial tests were conducted to evaluate mechanical behavior, while scanning electron microscopy (SEM) combined with the Pore and Crack Analysis System (PCAS) was used to quantify microstructural evolution. Results indicated that increasing FT cycles transformed the stress–strain response from mild strain-softening to strain-hardening, with the failure mode evolving toward bulging-type ductile failure. Cohesion exhibited a pronounced exponential decay, with the most significant degradation occurring within the first three FT cycles and stabilizing after approximately six FT cycles, whereas the internal friction angle showed only minor variation. At the microscale, porosity and total pore area increased continuously through micropore coalescence and macropore development, with a slight decrease in fractal dimension indicating reduced pore boundary complexity and smoothed pore interfaces due to frost heave-induced pore merging. The FT-induced hydrothermal disturbance promoted pore-water phase transition and redistribution, resulting in progressive pore enlargement and loss of structural integrity. Because the specimens were tested in sealed, closed-system conditions with a nearly constant total water content, this degradation chain is attributable specifically to in situ ice–water phase transitions and internal pore-water redistribution, i.e., water-phase-change-driven processes, rather than to external water supply. It is demonstrated that interparticle bond breakage and pore expansion–coalescence driven by ice–water phase transitions dominate strength degradation, promoting a transition from structure-dominated to friction-dominated strength behavior. A normalized cohesion reduction factor and a cohesion degradation index are further proposed to quantify the progressive loss of structural integrity and to provide a design-oriented tool for cold-region geotechnical practice. These findings provide a basis for stability assessment and hazard mitigation of dispersive soils in cold-region engineering.
In cold regions, solar radiation triggers the spring ablation of river ice layers, thereby changing their physical traits and mechanical behavior. This study uses the Heilongjiang River section near Mohe Arctic Village as the research prototype area. It analyzes the impact of solar radiation on ice density and uniaxial compressive strength through indoor simulation tests and multiple regression analysis, aiming to reveal the influence mechanism on uniaxial compressive strength. The results show that after applying a cumulative amount of simulated solar radiation of 84 MJ/m2, the ice density decreases by 3.88%, and the loss rate of uniaxial compressive strength can exceed 50%. Solar radiation promotes the transformation of the uniaxial compressive failure mode from ductile to brittle. The established multiple regression model attains a coefficient of determination of 0.891. In the spring ice-melting period in cold regions, the impact of solar radiation on ice strength should be fully considered in the design of ice condition early warnings and water conservancy projects for ice flood prevention.