Hysteretic curve characteristics in rectangular shear walls predicted by machine learning.

Area of Science:

• Structural Engineering
• Earthquake Engineering
• Machine Learning Applications

Background:

• Rectangular reinforced concrete (RC) shear walls are vital for seismic resistance in high-rise structures.
• Accurate prediction of characteristic points on the skeleton curve is essential for seismic performance evaluation.
• Existing models face challenges in capturing complex, nonlinear relationships between component dimensions and performance points.

Purpose of the Study:

• To develop an interpretable empirical guidance machine learning (IEG-ML) model for predicting key seismic performance feature points in RC shear walls.
• To enhance the accuracy and efficiency of seismic evaluation through an explainable AI approach.
• To identify critical component factors influencing seismic behavior.

Main Methods:

• Development of an interpretable empirical guidance machine learning (IEG-ML) model.
• Training the model on a self-built dataset of 184 reinforced concrete shear wall samples.
• Optimization of the IEG-ML model using a population optimization algorithm, specifically the dung beetle algorithm (DBO).
• Analysis of feature point importance and component factor influence.

Main Results:

• The IEG-ML model demonstrated high accuracy and efficiency in predicting seismic feature points.
• The dung beetle algorithm (DBO) optimization significantly enhanced model performance.
• The model successfully identified dominant component factors and provided an explicable formula for seismic evaluation.
• IEG-ML proved to be a robust tool for seismic performance assessment.

Conclusions:

• The proposed IEG-ML model offers a reliable and explainable method for predicting seismic performance of RC shear walls.
• Optimization with algorithms like DBO is crucial for maximizing the accuracy and efficiency of such models.
• This approach provides valuable insights for structural engineers in designing seismically resilient high-rise buildings.