Prediction of Compressive Strength of High-Performance Concrete: Hybrid Artificial Intelligence Technique

Abstract

Compressive strength is the most important mechanical property of concrete due to its significant role in numerous design codes and standards. Precise and early estimation of compressive strength of concrete can reduce cost and save time. Many studies have demonstrated that the development of concrete strength is determined not only by the water-to-cement ratio, nevertheless, that is also affected by the content of other concrete parameters and ingredients. High-performance concrete (HPC) is considered an extremely complicated material and the modelling of its performance and behavior is extremely difficult. In this study, Multiple Linear Regression (MLR) and Artificial Neural Networks (ANN) approaches coupled with cross validation technique (CV) used to predict the compressive strength of (HPC). The result showed that ANN-CV model has a good agreement between experimental and predicted compressive strength of concrete values compared to MLR model. The performance of ANN-CV model in estimation compressive strength was very well and superior to MLR-CV model base on statistical criteria such as Correlation Coefficient (CC), Root Mean Square Error (RMSE) and Coefficient of Residual Mass (CRM). The outcomes of this study also revealed that proposed ANN-CV model preforms better than MLR-CV model with higher value of correlation and fewer error was noticed (CC = 0.965, RMSE = 4.736 (MPA), CRM = −0.019) compared to MLR-CV (CC = 0.789, RMSE = 8.288 (MPA), CRM = 0.008).