Gaussian Regression Process for Prediction of Compressive Strength of Thermally Activated Geopolymer Mortars

Abstract

The primary objective of this research is the development of a prediction model of the compressive strength of geopolymer mortars made with fly ash and granular slag which hardened in different curing conditions. Data for the numerical analysis were obtained by experimental research; for this purpose 45 series of geopolymer mortars were made, 9 of which were cured in ambient conditions at a temperature of 22 °С, and the remaining were exposed to thermal activation for a duration of 24 h at the temperatures of 65 °С, 75 °С, 85 °С and 95 °С. Using machine learning, a Gaussian regression method was developed in which the curing temperature and the percentage mass content of fly ash and granular slag were used as input parameters, and the compressive strength as the output. Based on the results of the developed model, it can be concluded that the Gaussian regression process can be used as a reliable regression method for predicting the compressive strength of geopolymer mortars based on fly ash and granular slag.

Keywords

• curing temperature
• fly ash
• Gaussian regression process
• geopolymer mortars
• ground granulated blast furnace slag
• machine learning