The dynamic resilient modulus (MR) of a subgrade soil is a fundamental parameter for evaluating the dynamic stability and service resilience of subgrade fillers and structures, as well as an instrumental input for calculating the mechanical response and fatigue life of a pavement structure. To accurately and reasonably characterise the MR of subgrade soils, machine learning (ML) models were established using the support vector machine, random forest, and extreme gradient boosting algorithms based on a large-scale dataset including 3533 records of MR tests conducted on subgrade soils. Meanwhile, the weighted plasticity index (WPI), initial moisture content (w), dry unit weight (gamma d), confining stress (sigma c), deviator stress (sigma d), and numbers of freeze-thaw cycles (NFT) were set as the input variables to predict the MR using ML models, which considered the effects of wheel loads, physical properties and climate fluctuation on the subgrade soils during the service period. Subsequently, the Shapley additive explanations method was developed to explain the prediction model for the MR of subgrade soils based on ML algorithms. The results quantitatively illustrated the explicit mapping relationship and internal influencing mechanism between the significant features of the influences and MR of subgrade soils, which was consistent with prior experimental and physical cognition. In summary, the study findings provide meaningful guidelines for the structural design and life evaluation of pavement subgrade engineering.

Liquefaction is a significant challenge in the fields of earthquake risk assessment and soil dynamics, as it has the potential to cause extensive damage to buildings and infrastructure through ground failure. During the 2011 Great East Japan Earthquake, Urayasu City in the Chiba Prefecture experienced severe soil liquefaction, leading to evacuation losses due to the effect of the liquefaction on roads. Therefore, developing quantitative predictions of ground subsidence caused by liquefaction and understanding its contributing factors are imperative in preparing for potential future mega-earthquakes. This research is novel because previous research primarily focused on developing predictive models for determining the presence or absence of liquefaction, and there are few examples available of quantitative liquefaction magnitude after liquefaction has occurred. This research study extracts features from existing datasets and builds a predictive model, supplemented by factor analysis. Using the Cabinet Office of Japan's Nankai Trough Megathrust Earthquake model, liquefaction-induced ground subsidence was designated as the dependent variable. A gradient-boosted decision-tree (GDBT) prediction model was then developed. Additionally, the Shapley additive explanations (SHAP) method was employed to analyze the contribution of each feature to the prediction results. The study found that the XGBoost model outperformed the LightGBM model in terms of predictive accuracy, with the predicted values closely aligned with the actual measurements, thereby proving its effectiveness in predicting ground subsidence due to liquefaction. Furthermore, it was demonstrated that liquefaction assessments, which were previously challenging, can now be interpreted using SHAP factors. This enables accountable wide-area prediction of liquefaction-induced ground subsidence.