Evaluation of high performance concrete hardness properties using fuzzy logic based modeling.

Area of Science:

• Construction Engineering
• Materials Science
• Artificial Intelligence

Background:

• Accurate prediction of high-performance concrete (HPC) properties (compressive strength, slump) is crucial for effective construction engineering.
• Complex, non-linear relationships and data variability in factors like water-cement ratio and additives challenge traditional modeling approaches.
• Existing methods often fail to adequately address the uncertainties inherent in predicting concrete performance.

Purpose of the Study:

• To introduce a novel Hybrid Fuzzy Artificial Neural Network (HFANN) architecture for enhanced prediction of HPC properties.
• To integrate fuzzy logic with machine learning models (SVR, MLP, GBM) optimized by Chaos Game Optimization (CGO).
• To improve the accuracy and robustness of predictive models for HPC mix design and sustainable construction.

Main Methods:

• Developed a Hybrid Fuzzy Artificial Neural Network (HFANN) integrating fuzzy logic (Gaussian membership functions, Sugeno-type rules) with SVR, MLP, and GBM.
• Utilized Chaos Game Optimization (CGO) for hyperparameter and ensemble weight optimization via chaotic search.
• Preprocessed inputs using fuzzy logic to generate 18 fuzzy features per sample, capturing data uncertainties.

Main Results:

• The HFANN framework demonstrated superior predictive accuracy, achieving a 25% reduction in Root Mean Square Error (RMSE) for compressive strength.
• Achieved high R² values: 0.98 for compressive strength and 0.97-0.99 for slump across training, validation, and testing datasets.
• MANOVA and Tukey's HSD tests confirmed significant improvements over conventional predictive methods.

Conclusions:

• The proposed HFANN framework provides a robust and accurate tool for predicting critical HPC properties.
• This approach effectively handles complex relationships and uncertainties in concrete mix design.
• The HFANN model advances sustainable construction practices through improved HPC mix design efficiency.