Novel genetic algorithm for optimizing Geopolymer concrete and predicting compressive strength

Abstract The rising demand for cement and natural aggregates in construction has led to the depletion of renewable resources, highlighting the need for sustainable alternatives. Geopolymer‐concrete (GPC), produced from industrial by‐products like fly ash and ground granulated blast‐furnace slag (GGBS), offers an eco‐friendly substitute for traditional cement‐based concrete. However, developing an effective GPC mix is complex due to the interdependence of factors influencing strength and durability. This study focuses on optimizing GPC mix designs using correlation analysis, scatter plots, multi‐objective optimization, and predictive models such as support vector regression, decision tree, random forest, and genetic algorithm (GA). The correlation matrix results suggest the influence of GGBS and fly ash (FA) on compressive strength in GPC, with strong positive correlations of 0.96 and 0.97, respectively. Similarly, coarse aggregate exhibited a strong negative correlation of 0.97. Compressive strength of GPC was increased by the ideal dosages of NaOH, Na₂SiO₃, GGBS, and fly ash, whereas it was decreased by excessive alkalinity or material imbalances. To attain the maximum compressive strength of 47 MPa, multi‐objective optimization determined the ideal ranges for GGBS (100–110 kg), fly ash (200 kg), and coarse aggregate (400 kg), highlighting the significance of exact material balance. These results were further validated using a regression model optimized by GA, which outperformed SVR, decision tree, and random forest models, achieving the best prediction accuracy with an R 2 of 0.97 and an RMSE of 1.27. The reliability of the model was further ensured through fivefold cross‐validation, which helped prevent overfitting and confirmed the generalization capability of the GA‐based model.