Prediction Of Strength Properties Of Geopolymer Concrete By Using Artificial Intelligence And Machine Learning Techniques

Abstract

Geopolymer concrete (GPC) is a feasible option in contrast to customary cement, utilizing fly debris (FA) rather than conventional Portland concrete (OPC). This study utilized quality articulation programming (GEP) and multi-articulation programming (MEP) to foster models anticipating compressive strength (CS) and split rigidity (ST) of FA-based GPC. A data set of 301 CS and 96 ST results, with seven information factors (FA, sodium hydroxide, sodium silicate, water, superplasticizer, fine and coarse totals), was utilized. The GEP-based models beat MEP-based models, accomplishing connection coefficients (R) of 0.89 for CS and 0.87 for ST, contrasted with 0.76 and 0.73 for MEP models. Mean outright mistakes were 5.09 MPa (CS) and 0.42 MPa (ST) for GEP, and 6.78 MPa (CS) and 0.51 MPa (ST) for MEP. GEP-based models offer straightforward numerical definitions and pragmatic application potential in enhancing GPC blend plans, supporting reasonable development rehearses.

geopolymer concrete; compressive strength; split tensile strength; prediction model; evolutionary algorithm