Abstract
•Machine learning models were used to predict strength of bagasse ash concrete.•Waste sugarcane bagasse ash (SCBA) was processed, characterized and optimized.•A number of concrete samples were prepared in laboratory to validate the ML model.•Robust models were developed using integration of literature and experimental data.
The present study aimed to develop models for estimation of the compressive strength (f’c) of sugarcane bagasse ash (SCBA) concrete through experimental testing and three machine learning (ML) approaches, namely gene expression programming (GEP), random forest regression (RFR) and support vector machine (SVM). The models were calibrated based on a widespread literature dataset of five input variables i.e., SCBA dosage (SCBA%), the quantity of fine aggregate (FA) and coarse aggregate (CA), water-cement ratio (W/C) and cement content (CC). The performance of each model was evaluated using relative squared error (RSE), root mean square logarithmic error (RMSLE), root mean squared error (RMSE), mean absolute error (MAE), Nash-Sutcliffe efficiency (NSE), percentage of relative root mean squared error (RRMSE%), coefficient of determination (R2) and performance index (PI). The suggested models prediction was checked and validated against the actual dataset acquired from laboratory tested SCBA concrete. The comparative study of the models revealed that RFR is an effective approach providing a strong correlation between actual and estimated outcome. The R2 and NSE for all the models are above 0.85 each with RRMSE% and PI less than 10% and 0.2, respectively. The GEP leads in providing a simple and reliable mathematical expression for estimation of f’c of SCBA concrete. The sensitivity analysis reflected the increasing order of the influence of input variables followed the trend: CC (55.73%) > W/C (17.15%) > CA (16.38%) > SCBA% (6.38%) > FA (3.76%) and are in close agreement with the experimental study conducted. Eachproposed ML approachoutperform on the literature data as well as dataset from laboratory tested SCBA concrete and also having superior generalization and prediction capacity. Conclusively, the resultsof this research can enable practitioners,researchers, and designersin quickly evaluating the f'c of SCBA concrete, thereby reducingenvironmental susceptibilities and resulting in safer, faster, and more sustainable construction from the perspective of eco-friendly waste management.