Estimating compressive strength of lightweight foamed concrete using neural, genetic and ensemble machine learning approaches

Foamed concrete is special not only in terms of its unique properties, but also challenging compositional mixture design, which necessitates multiple experimental trials before obtaining the desired property like compressive strength. Regardless design challenges, artificial intelligence (AI) techniques have shown to be useful reliably estimating properties based on optimized proportions. This study proposes AI-based models predict strength foamed concrete. Three novel AI approaches, namely neural network (ANN), gene expression programming (GEP), and gradient boosting tree (GBT) models, were employed. The developed using 232 results, considering easily acquired variables, such as density concrete, water-cement ratio sand-cement inputs estimate In training 80% data was used rest validate models. selected their respective best hyper-parameters trial error basis; variable number hidden layers, neurons algorithms for ANN, chromosomes, head size, genes, function set GEP GBT employed trees, maximal depth learning rate. trained validated parametric sensitivity analyses a simulated dataset. prediction abilities proposed evaluated coefficient correlation (R), mean absolute (MAE), root squared (RMSE). For validation data, empirical results from performance evaluation revealed that model (R = 0.977, MAE 1.817 RMSE 2.69) has relative superior with highest least comparison ANN 0.975, 2.695 3.40) 0.96, 2.07 2.80). concludes offered reliable accuracy predicting Finally, simple equation generated signifies importance can It recommended shall ranges input variables this study.