Predicting the Compressive Strength of Sustainable Portland Cement-Fly Ash Mortar Using Explainable Boosting Machine...

Area of Science:

• Materials Science
• Civil Engineering
• Artificial Intelligence

Background:

• Unconfined compressive strength (UCS) is vital for cement-fly ash mortar (CFAM) in construction.
• Experimental UCS determination is costly and time-consuming.
• Machine learning offers a potential solution for efficient UCS prediction.

Purpose of the Study:

• To model and predict the UCS of CFAM using boosting machine learning methods.
• To identify the most influential input parameters affecting CFAM's UCS.
• To compare the performance of different machine learning models for UCS prediction.

Main Methods:

• Developed a database of 395 experimental CFAM data points.
• Employed Gradient Boosting Regressor (GBR), Light Gradient Boosting Machine (LGBM), and Ada-Boost Regressor (ABR) models.
• Utilized SHapley Additive exPlanations (SHAP) for feature importance analysis.

Main Results:

• The GBR model demonstrated superior accuracy in predicting CFAM's UCS compared to LGBM and ABR.
• SHAP analysis identified curing time as the most critical factor, followed by fly ash's Al2O3 content.
• Predicted UCS values showed no significant difference from measured values, indicating high model reliability.

Conclusions:

• Boosting machine learning, particularly GBR with SHAP, is effective for modeling CFAM's UCS.
• This approach can significantly reduce the time and cost associated with experimental testing.
• The findings support the use of advanced computational methods in sustainable material design and engineering.