Machine learning models development for shear strength prediction of reinforced concrete beam: a comparative study

Area of Science:

• Civil Engineering
• Materials Science
• Computational Mechanics

Background:

• Fiber Reinforced Polymer (FRP) bars are increasingly used as a substitute for steel reinforcement in concrete structures, especially in corrosive environments.
• Understanding the shear strength (Vs) of FRP-reinforced concrete elements is crucial for accurate pre-design, as it is influenced by concrete properties and transverse FRP stirrups.

Purpose of the Study:

• To evaluate the predictive capabilities of three machine learning (ML) models—M5-Tree (M5), Extreme Learning Machine (ELM), and Random Forest (RF)—for estimating the shear strength (Vs) of FRP-reinforced concrete beams.
• To identify the most effective ML model and input parameters for predicting Vs in FRP-reinforced concrete beams with transverse reinforcement.

Main Methods:

• Collected data from 112 shear tests on FRP-reinforced concrete beams with transverse reinforcement.
• Employed statistical correlation analysis to determine optimal input parameters for the ML models.
• Developed and evaluated M5-Tree, ELM, and RF models to predict shear strength (Vs).
• Utilized statistical evaluation and graphical approaches to assess model performance.

Main Results:

• All investigated ML models generally performed well in predicting shear strength (Vs).
• The M5-Tree model, using nine input parameters, achieved the highest prediction accuracy with R² = 0.9313 and RMSE = 35.5083 KN.
• ELM and RF models also yielded significant results, with slightly lower performance compared to the optimal M5-Tree model.
• Specific models like ELM-M1 and M5-Tree-M5 showed comparatively lower accuracy.

Conclusions:

• Machine learning models, particularly M5-Tree, are effective tools for predicting the shear strength of FRP-reinforced concrete beams.
• The study highlights the significant impact of transverse reinforcement on the shear strength of FRP concrete beams.
• The findings contribute to the fundamental understanding of shear mechanisms and the application of computational models in structural engineering design.