Scholarlink

<p>Spalling of the longitudinal joints of Continuously Reinforced Concrete Pavement can be vital to successful pavement management and efficient maintenance cost planning. An optimized predictive model is established by integrating the Genetic Algorithm with the Gradient Boosting Machine in such a manner to enhance the level of accuracy of the model. A curated dataset of 395 observations from the Long-Term Pavement Performance (LTPP) database, comprising 20 structural, climatic, and traffic-related variables, was used. The GA-GBM model’s performance was benchmarked against standard GBM, Linear Regression, Random Forest, and Artificial Neural Networks (ANN). The proposed GA-GBM achieved the highest accuracy, with a mean Root Mean Square Error (RMSE) of 12.916 and a mean coefficient of determination (R²) of 0.945, surpassing all comparative models. Sensitivity analysis revealed that Age, Kilo Equivalent Single Axle Loads (KESAL), and Annual Average Daily Traffic (AADT) are the most influential predictors of spalling progression. Heatmaps, residual plots, and comparisons of predicted and measured values were used to assess model robustness. The findings demonstrate the potential of combining GA optimization with ensemble learning to improve distress prediction, providing transportation agencies with a dependable tool for prioritizing interventions and improving pavement management systems.</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

Advanced prediction of spalling in rigid pavements using GBM and GA optimization

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年9月26日

查看全文
  • Ali Alnaqbi,
  • Ghazi G. Al-Khateeb,
  • Waleed Zeiada

摘要

Spalling of the longitudinal joints of Continuously Reinforced Concrete Pavement can be vital to successful pavement management and efficient maintenance cost planning. An optimized predictive model is established by integrating the Genetic Algorithm with the Gradient Boosting Machine in such a manner to enhance the level of accuracy of the model. A curated dataset of 395 observations from the Long-Term Pavement Performance (LTPP) database, comprising 20 structural, climatic, and traffic-related variables, was used. The GA-GBM model’s performance was benchmarked against standard GBM, Linear Regression, Random Forest, and Artificial Neural Networks (ANN). The proposed GA-GBM achieved the highest accuracy, with a mean Root Mean Square Error (RMSE) of 12.916 and a mean coefficient of determination (R²) of 0.945, surpassing all comparative models. Sensitivity analysis revealed that Age, Kilo Equivalent Single Axle Loads (KESAL), and Annual Average Daily Traffic (AADT) are the most influential predictors of spalling progression. Heatmaps, residual plots, and comparisons of predicted and measured values were used to assess model robustness. The findings demonstrate the potential of combining GA optimization with ensemble learning to improve distress prediction, providing transportation agencies with a dependable tool for prioritizing interventions and improving pavement management systems.