Prediction of Compressive Strength of Alccofine-Based Geopolymer Concrete: Fid-Bau Portal

Prediction of Compressive Strength of Alccofine-Based Geopolymer Concrete

Diksha / Dev, Nirendra / Goyal, Pradeep Kumar

Geopolymer concrete has been proposed as a superior substitute for ordinary Portland cement concrete, owing to its superior performance under harsh conditions. It significantly decreases the carbon footprint to a considerable extent. The utilization of machine learning techniques in the construction industry is poised to become increasingly prevalent due to its ability to forecast the mechanical properties of concrete mix designs based on their components, thereby obviating the need for destructive testing. The objective of this research is to construct predictive models for the compressive strength of geopolymer concrete mixed with alccofine and subsequently verify their accuracy through comparison with observed values. The results of the experimental investigation indicate that the compressive strength is greater when subjected to oven-dried curing in comparison to ambient curing conditions. The addition of alccofine results in an increase in compressive strength up to a certain percentage, beyond which a decrease is observed. Machine learning techniques such as support vector regression (SVR), gene expression programming (GEP), and artificial neural network (ANN) have been deployed to construct models based on empirical observations. The output variable considered in this study is compressive strength, while the input variables comprise a range of influencing parameters that are utilized for both training and testing purposes. The efficacy of anticipated values is evaluated through diverse statistical parameters. The ANN model outperforms the SVR and GEP models in terms of predictive accuracy, as evidenced by its superior statistical performance when comparing actual and predicted values.