Implementation of nonlinear computing models and classical regression for predicting compressive strength of high-performance concrete

Abstract

The construction sector would greatly benefit from a strategy for optimizing high-performance concrete mixtures. However, traditional proportioning techniques are insufficient because of their high prices, usage restrictions, and inability to account for nonlinear interactions between components and concrete qualities. High-performance concrete (HPC) is a complicated composite material with highly nonlinear mechanical behaviour. When strength can be accurately predicted, design costs, design time, and material waste caused by several mixing trials can all be reduced. In this research, feed-forward neural network (FFNN), Elman neural network (ENN), support vector machine (SVM) and multilinear regression (MLR) were employed for predicting the compressive strength of HPC. The input variables include cement (C), cement strength (CeS), superplasticizer (S), fly ash (F), air entraining agent (A), coarse aggregate (CA), Sand (Sd) and water/binder (W/B) and 28 days’ compressive strength as the output variables. Finally, the results indicate that the proposed model has predictive robustness for predicting the compressive strength of HPC. The results showed that FFNN-M4, ENN-M4, SVM-M4, and MLR-M4 combination have the highest performance evaluation criteria of R²=0.9950, R²=0.9853, R²=0.9736, R²= 0.9678 in the testing phase respectively. The outcomes also show that the proposed model has high accuracy and effectiveness in predicting the compressive strength of HPC.