Abstract
This study employed an AI-based strategy to determine the accurate parameters in alkali-activated concrete (AAC) mix design that contribute to optimal performance. The data mining approach was used to generate a dataset comprising seven crucial input parameters and three outputs. Gene expression programming (GEP) models and multi-expression programming (MEP) models were developed to predict the rheological characteristics (in terms of static/dynamic yield stress and plastic viscosity) of AAC. R2 values, statistical checks, Taylor's diagram, and the difference between experimental and anticipated rheological parameters were used to assess the suitability of the created models. All prediction models developed using the MEP approach were found to be highly accurate (R2 > 0.90), while all GEP models were found to be in the acceptable range of accuracy (R2 near 0.90). Moreover, the greater accuracy of MEP models over GEP models was also confirmed by error evaluation by statistical tests. The empirical equations presented by the models might be useful for comprehending the mix design of AAC and the effect of each input parameter. Precursor content was found to have a significantly positive impact on ACC's rheological properties, as determined by the sensitivity analysis.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have