Predicting Compressive Strength of Concrete Containing Industrial Waste Materials: Novel and Hybrid Machine Learning Model

In the construction and cement manufacturing sectors, development of artificial intelligence models has received remarkable progress attention. This paper investigates capacity hybrid conducted for predicting compressive strength (CS) concrete where was partially replaced with ground granulated blast-furnace slag ( FS ) fly ash id="M2"> FA materials. Accurate estimation CS can reduce cost laboratory tests. Since traditional method calculation is complicated requires lots effort, this article presents new predictive called id="M3"> SVR − PSO id="M4"> GA , that are a hybridization support vector regression id="M5"> improved particle swarm algorithm id="M6"> genetic id="M7"> ). Furthermore, (i.e., id="M8"> id="M9"> were used first time to predict component replaced. The id="M10"> id="M11"> given essential roles in tuning hyperparameters SVR model, which have significant influence on model accuracy. suggested evaluated against extreme learning machine (ELM) via quantitative visual evaluations. using eight statistical parameters, then SVR-PSO provided highest accuracy than comparative models. For instance, id="M12"> during testing phase fewer root mean square error id="M13"> RMSE 1.386 MPa, higher Nash–Sutcliffe efficiency coefficient id="M14"> NE 0.972, lower uncertainty at 95% id="M15"> U 95 28.776%. On other hand, id="M16"> id="M17"> ELM provide id="M18"> 2.826 MPa 2.180, id="M19"> 0.883 0.930, id="M20"> 518.686 183.182, respectively. Sensitivity analysis carried out select influential parameters significantly affect id="M21"> CS . Overall, proposed showed good prediction outperformed 14 developed previous studies.