Using Artificial Intelligence and Machine Learning Model for Prediction of Uniaxial Compressive Strength of GGBS Concrete

Abstract

Ground Granulated Blast-furnace Slag (GGBS) concrete is sustainable and also proven to show enhanced properties such as reduced heat of hydration, refinement of pore structure and increased resistance to corrosion and chemical attacks when compared with Ordinary Portland Cement concrete. Uniaxial compressive strength (UCS) and enhanced properties of the GGBS concrete are dependent on the proportions of GGBS and other elements in the mix. It is essential to accurately predict the UCS of the blended mix. Developments in computer hardware and the easy availability of research data made artificial intelligence and machine learning (AI & ML) prediction techniques feasible. In this study AI & ML techniques namely, linear regression, lasso regression and ridge regression are used to predict the UCS of GGBS concrete. Algorithms are trained using data collected from various standard publications. All data points are cleaned and then normalized with standard scalar function to avoid biased predictions. Root Mean Squared Error, Mean Squared Error, Mean Absolute Error and Coefficient of Determination for all three techniques are found to be almost identical. The above-mentioned AI & ML models have high prediction accuracy. Hence, based on the results AI & ML algorithms specified above can be reliably used in the above-mentioned GGBS concrete.