Sign in Subscribe
Data mining

Machine Learning Predicts Compressive and Flexural Strengths of Ultra-High-Performance Concrete

Strategic Insights

— 1 min read

Researchers from Shahrood University of Technology have utilized machine learning methods to predict the compressive and flexural strengths of ultra-high-performance concrete (UHPC) with various fiber types. The study employed six machine learning algorithms, including gradient boosting, light gradient-boosting machine, and deep neural networks, to analyze a dataset of 321 and 863 experimental data points for flexural and compressive strength, respectively. The results indicated that the categorical gradient boosting (CatBoost) algorithm was the most effective predictor of the model for both compressive and flexural strength, with R2 values of 0.9309 and 0.9210 for the test data.

Key Takeaways:

  • The study used machine learning methods to predict the compressive and flexural strengths of UHPC with various fiber types, including steel, basalt, glass, and polypropylene fibers.
  • The researchers employed six machine learning algorithms, including gradient boosting, light gradient-boosting machine, and deep neural networks, to analyze the dataset.
  • The study considered a total of 321 and 863 experimental data points for flexural and compressive strength, respectively.
  • The results of the study were interpreted using shapley additive explanations (SHAP), a technique used to explain the predictions of machine learning models.
  • The researchers developed a graphical user interface (GUI) to make the written program more accessible and user-friendly.
  • The study's conclusions are valuable for construction applications and provide designers and builders with practical insights into the elements of UHPC.
  • The source codes and data files are available on GitHub at https://github.com/MiladBolbolvand/UHPC-Machine-Learning.git.

Statistics:

  • 321 experimental data points were used to analyze flexural strength.
  • 863 experimental data points were used to analyze compressive strength.
  • The R2 values of the test data for compressive strength were 0.9309 and 0.9210, respectively, for the CatBoost algorithm.
  • The study employed six machine learning algorithms, including gradient boosting, light gradient-boosting machine, and deep neural networks.

Sources:

  • NewsRx. Findings from Shahrood University of Technology in the Area of Machine Learning Described [Prediction of Compressive and Flexural Strengths of Ultra-high-performance Concrete (Uhpc) Using Machine Learning for Various Fiber Types]. Journal of Engineering. October 20, 2025; p 745.
  • Shahrood University of Technology. Prediction of Compressive and Flexural Strengths of Ultra-high-performance Concrete (Uhpc) Using Machine Learning for Various Fiber Types. Construction and Building Materials, 2025;493.

Read more