Abstract—Accurate prediction of the compressive strength of High-Performance Concrete (HPC) is crucial in concrete design and construction. However, HPC is a very complex material, as the inter-relationship between its constituent materials is highly nonlinear and its property is affected by several interacting factors. Hence, existing conventional empirical and statistical methods are limited in their ability to accurately predict the compressive strength of HPC. In this study, the application of three artificial intelligence techniques, namely, the Artificial Neural Network (ANN), Fuzzy Inference System (FIS), and Adaptive Neuro-Fuzzy Inference System (ANFIS) techniques, are explored. A data-driven approach based on fuzzy c-means clustering (FCM) is employed to generate both the Mamdani and Sugeno FIS models. Different model structures and parameters—such as number of neurons and choice of transfer function for the ANN technique, and number of clusters and choice of fuzzification coefficient and inference methods for the FIS and ANFIS techniques—are optimized to improve the accuracy of each technique. Results of this study indicate that ANFIS and ANN perform better than the FIS models in predicting the compressive strength of HPC. The main contributions of this paper are: (1) providing accurate concrete compressive strength prediction models that represent the complex, nonlinear relationship between the constituent materials and concrete compressive strength; (2) presenting a data-driven methodology for the development of FIS concrete compressive strength models; and (3) subjecting artificial intelligence-based concrete compressive strength models to structure and parameter optimization to improve prediction accuracy.

Index Terms—high-performance concrete, compressive strength, artificial neural network, fuzzy inference system, adaptive neuro-fuzzy inference system

Cite: Nasir B. Siraj, Aminah Robinson Fayek, and Abraham A. Tsehayae, "Development and Optimization of Artificial Intelligence-Based Concrete Compressive Strength Predictive Models," International Journal of Structural and Civil Engineering Research, Vol. 5, No. 3, pp. 156-167, August 2016. doi: 10.18178/ijscer.5.3.156-167