Prediction Of Properties Of Geopolymer Concrete Using Experimental And Machine Learning Technique

Abstract

he reason for global warming is due to the emission of greenhouse gases from the newlinemanufacture of cement. Fly ash (FA) and Ground granulated blast furnace slag (GGBS) used newlineas binders to prepare geopolymer concrete (GPC) as an alternative in place of ordinary newlineconcrete for sustainable development. Compressive strength is the most crucial mechanical newlinecharacteristic of concrete for the quality assurance of engineering structures. This study newlinefocused on two primary precursor materials such as FA and GGBS in preparation of newlinegeopolymer concrete. newlineThe first phase of this study investigated the compressive strength of GGBS-based GPC by newlineusing both experimental and machine learning methods. The machine learning (ML) newlinetechnique is based on the experimental data from previous research and current experimental newlineresults. Then after, the generated database is used for analysis through various machine- newlinelearning models to check the accuracy of the prediction. To create a dataset for training and newlinetesting of the model, several experiments using various mix designs were carried out. The newlinecompressive strength of GGBS-based GPC is predicted in this study using a novel enhanced newlinecat swarm-optimized extreme learning machine (ELM-ECSO) model. The ELM-ECSO is newlinecompared with simple ELM and ELM with a cat swarm optimized to check its newlinegeneralizability. The quasi-Monte Carlo method was implemented for the sensitivity analysis newlineat the model. Assessment results showed that the ELM-ECSO model transcends the other two newlinemodels owing to the highest prediction accuracy and showing the least error. Shapiro-Wilk newlinestatistical test was carried out to compare all the models. The sensitivity analysis indicated newlinetemperature exposure curing age as the most impactful parameter while predicting the newlineresidual compressive strength of GPC. The second phase of this study focuses on newlineinvestigating the compressive strength of FA-based GPC using experimental investigation newlinevi newlineand the influence of chemical constituents in binders and activators in predicting newlineC