Prediction of concrete compressive strength using artificial neural network with bayesian optimization

Concrete is the most widely used material in the field of civil engineering, so it is crucial to determine the compressive strength of concrete. However, many studies rely on traditional methods such as experiments or numerical simulations, which have the disadvantage of being time-consuming and expensive. The goal of this study is to predict the compressive strength of concrete using an artificial neural network (ANN) algorithm combined with Bayesian Optimization (ANN-BO). For a comparative study with a proposed model, 10 other machine learning (ML) models, including Linear Regression (LR), Ridge Regression, LASSO Regression, Elastic Net Regression, Polynomial Regression, Stochastic Gradient Descent, Decision Tree (DT), Random Forest Regression, Support Vector Machine, and Baseline ANN were performed. From the literature, a wide range of factors that impacted the concrete compressive strength were considered as input database applied in the models, such as cement, blast furnace slag, fly ash, water, superplasticizer, coarse aggregate, fine aggregate, and ot curing age. The testing results indicate that all models exhibited an accuracy of R 2 value of over 0.615 and the ANN-BO model outperformed other methods. It surpassed existing approaches with an R 2 value of 0.896, RMSE of 5.187, and MAE of 3.992. With high reliability and robustness, the proposed ANN-BO algorithm can be used as an efficient tool for engineers and researchers.