Scholarlink

<p>Assessing subgrade soil behavior under repeated traffic loads critically relies on the resilient modulus, a fundamental parameter in the Mechanical Empirical Pavement Design Guide (MEPDG) for flexible pavements. Conventionally, determining the resilient modulus requires conducting numerous triaxial stress tests, which is a convoluted and time-consuming process. The present work investigates an alternate methodology for predicting the resilient modulus using soil characteristics, including coarse and fine content, liquid limit, plasticity index, dry density, moisture content, confining stress, and deviator stress. A range of multivariate models, including Linear Regression (LR), Multilinear Regression (MLR), Nonlinear Regression (NLR), Pure Quadratic (PQ), Interaction (IA), and Full Quadratic (FQ) models, were used in the study. The statistical evaluations, including the coefficient of determination (R<sup>2</sup>), scatter index (SI), root mean squared error (RMSE), mean absolute error (MAE), and Objective (OBJ) value, indicated that the Full Quadratic and Interaction models yielded the most precise predictions. In sensitivity analysis, moisture content was identified as a primary determinant of resilient modulus. By comparison, the predictive accuracy of Nonlinear and Multilinear regression models in estimating resilient modulus was lower.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Data-driven prediction of resilient modulus in flexible pavement subgrades using soil properties and stress variables

  • 期刊论文

  • 发表日期：

    2026年4月4日

查看全文
  • Hunar Farid Hama Ali,
  • Ahmed Salih Mohammed

摘要

Assessing subgrade soil behavior under repeated traffic loads critically relies on the resilient modulus, a fundamental parameter in the Mechanical Empirical Pavement Design Guide (MEPDG) for flexible pavements. Conventionally, determining the resilient modulus requires conducting numerous triaxial stress tests, which is a convoluted and time-consuming process. The present work investigates an alternate methodology for predicting the resilient modulus using soil characteristics, including coarse and fine content, liquid limit, plasticity index, dry density, moisture content, confining stress, and deviator stress. A range of multivariate models, including Linear Regression (LR), Multilinear Regression (MLR), Nonlinear Regression (NLR), Pure Quadratic (PQ), Interaction (IA), and Full Quadratic (FQ) models, were used in the study. The statistical evaluations, including the coefficient of determination (R2), scatter index (SI), root mean squared error (RMSE), mean absolute error (MAE), and Objective (OBJ) value, indicated that the Full Quadratic and Interaction models yielded the most precise predictions. In sensitivity analysis, moisture content was identified as a primary determinant of resilient modulus. By comparison, the predictive accuracy of Nonlinear and Multilinear regression models in estimating resilient modulus was lower.