Scholarlink

<p>Permanent deformation is important distress mechanism in hot mix asphalt, resulting in premature pavement failure under high temperatures condition and heavy traffic loads. This study proposes Genetic Algorithm models to estimate hot mix asphalt permanent deformation using a database of 118 laboratory-prepared asphalt mixtures. Input parameters comprised coarse and fine aggregate contents (Coarse% = 42–81%, Fine% = 18–48%), bitumen content 4.0–6.0%, air voids = 1.71–8.77%, voids in mineral aggregate = 13.2–18.29%, Marshall Stability (2.73–15.30 kN), and flow (2.0–4.75&#xa0;mm). Flow number, ranging from 22 to 510 cycles, was adopted as the rutting performance indicator and produced from dynamic creep testing. Five Genetic Algorithm models were proposed in MATLAB, achieving coefficients of determination (R² = 0.92–0.93) for both training and testing datasets. The most accurate models demonstrated that aggregate gradation and bitumen content alone predict Flow number with R² = 0.93, indicating their dominant influence on rutting resistance. Mixtures with higher fine aggregate content and natural fines exhibited up to 60% reduction in Flow number, indicating significantly increased deformation susceptibility. The robustness of the model was analyzed using five-fold cross-validation, which demonstrated a mean R<sup>2</sup> value of 0.91 and a low percentage of variability (standard deviation = 0.022). This demonstrated that the model had a strong generalization capability and minimal overfitting. As an additional point of interest, sensitivity analysis based on Pearson correlation demonstrated that aggregate gradation parameters collectively contribute around 59% to Flow number prediction, with fine aggregate content being identified as the most important element.</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

Introducing mathematical modeling to estimate permanent deformation of hot mix asphalt using genetic algorithm

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月1日

查看全文
  • Shadi M. Hanandeh,
  • Frank I. Aneke,
  • Ahamad Hanandeh,
  • Saeed Alzahrani,
  • Rami Hanandeh,
  • Ruba A. Alkharabsheh

摘要

Permanent deformation is important distress mechanism in hot mix asphalt, resulting in premature pavement failure under high temperatures condition and heavy traffic loads. This study proposes Genetic Algorithm models to estimate hot mix asphalt permanent deformation using a database of 118 laboratory-prepared asphalt mixtures. Input parameters comprised coarse and fine aggregate contents (Coarse% = 42–81%, Fine% = 18–48%), bitumen content 4.0–6.0%, air voids = 1.71–8.77%, voids in mineral aggregate = 13.2–18.29%, Marshall Stability (2.73–15.30 kN), and flow (2.0–4.75 mm). Flow number, ranging from 22 to 510 cycles, was adopted as the rutting performance indicator and produced from dynamic creep testing. Five Genetic Algorithm models were proposed in MATLAB, achieving coefficients of determination (R² = 0.92–0.93) for both training and testing datasets. The most accurate models demonstrated that aggregate gradation and bitumen content alone predict Flow number with R² = 0.93, indicating their dominant influence on rutting resistance. Mixtures with higher fine aggregate content and natural fines exhibited up to 60% reduction in Flow number, indicating significantly increased deformation susceptibility. The robustness of the model was analyzed using five-fold cross-validation, which demonstrated a mean R2 value of 0.91 and a low percentage of variability (standard deviation = 0.022). This demonstrated that the model had a strong generalization capability and minimal overfitting. As an additional point of interest, sensitivity analysis based on Pearson correlation demonstrated that aggregate gradation parameters collectively contribute around 59% to Flow number prediction, with fine aggregate content being identified as the most important element.