Scholarlink

<p>Concrete's split tensile strength (STS) is a crucial metric when assessing the material's structural integrity and longevity. The split tensile strength (STS) of concrete is a critical parameter for assessing its structural integrity and durability. Traditional methods for predicting STS involve labour-intensive testing procedures. This study applies advanced machine learning models, Gradient Boosting (GB), Random Forest (RF), and Adaptive Boosting (AdaBoost) to predict the STS of hollow concrete blocks (HCBs) based on the rod position during ASTM C-1006-13 split tensile testing. A dataset comprising 90 observations with 22 input parameters, including geometrical properties (block dimensions, cavity sizes, thicknesses) and experimental conditions (net area, applied load, block length, and height), was used for model training and evaluation. It enhanced predictive accuracy and address multicollinearity, Principal Component Analysis (PCA) was employed as a dimensionality reduction technique. The model’s performance was assessed using Root Mean Square Error (RMSE) and the coefficient of determination (R<sup>2</sup>). The Random Forest model demonstrated the highest accuracy, achieving RMSE = 0.118 and R<sup>2</sup> = 0.920 in the testing phase. Compared to conventional testing methods, the findings highlight the effectiveness of feature selection and machine learning techniques in developing reliable predictive models for concrete performance.</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

Predicting split tensile strength of hollow concrete blocks using PCA-enhanced machine learning models

  • 期刊论文

  • 发表日期：

    2025年6月10日

查看全文
  • S. Hetaish Subramanya,
  • S. Deepak Raj,
  • Rakesh Kumar,
  • Sathvik Sharath Chandra

摘要

Concrete's split tensile strength (STS) is a crucial metric when assessing the material's structural integrity and longevity. The split tensile strength (STS) of concrete is a critical parameter for assessing its structural integrity and durability. Traditional methods for predicting STS involve labour-intensive testing procedures. This study applies advanced machine learning models, Gradient Boosting (GB), Random Forest (RF), and Adaptive Boosting (AdaBoost) to predict the STS of hollow concrete blocks (HCBs) based on the rod position during ASTM C-1006-13 split tensile testing. A dataset comprising 90 observations with 22 input parameters, including geometrical properties (block dimensions, cavity sizes, thicknesses) and experimental conditions (net area, applied load, block length, and height), was used for model training and evaluation. It enhanced predictive accuracy and address multicollinearity, Principal Component Analysis (PCA) was employed as a dimensionality reduction technique. The model’s performance was assessed using Root Mean Square Error (RMSE) and the coefficient of determination (R2). The Random Forest model demonstrated the highest accuracy, achieving RMSE = 0.118 and R2 = 0.920 in the testing phase. Compared to conventional testing methods, the findings highlight the effectiveness of feature selection and machine learning techniques in developing reliable predictive models for concrete performance.