Concrete Strength Prediction Using Different Machine Learning Processes: Effect of Slag, Fly Ash and Superplasticizer

Area of Science:

• Materials Science
• Civil Engineering
• Computational Intelligence

Background:

• Blast furnace slag (BFS) and fly ash (FA) are industrial byproducts with pozzolanic properties.
• Utilizing BFS and FA in concrete can reduce cement content, promoting sustainable construction.
• Accurate prediction of concrete compressive strength is crucial for structural design and safety.

Purpose of the Study:

• To develop a robust prediction model for the compressive strength of concrete incorporating BFS and FA with superplasticizer.
• To optimize the model's hyperparameters using Particle Swarm Optimization (PSO).
• To evaluate the necessity of Principal Component Analysis (PCA) for feature reduction in this context.

Main Methods:

• A Random Forest (RF) model was employed for compressive strength prediction.
• Particle Swarm Optimization (PSO) was utilized for hyperparameter tuning of the RF model.
• Principal Component Analysis (PCA) was applied for dimensionality reduction of input features.

Main Results:

• The optimal RF-PSO model achieved high performance with R=0.954, EVS=0.901, MAE=3.746, and MSE=27.535 on the testing set.
• PCA dimensionality reduction slightly decreased the prediction accuracy (R=0.88), indicating it was not essential for this dataset.
• Sensitivity analysis identified cement as the most influential factor, followed by water, superplasticizer, fine aggregate, BFS, coarse aggregate, and FA.

Conclusions:

• The proposed RF-PSO model effectively predicts the compressive strength of BFS-FA-superplasticizer concrete.
• The study demonstrates the model's potential for practical engineering applications in sustainable concrete mix design.
• Understanding feature importance aids in optimizing concrete formulations for desired strength characteristics.