Predicting Failure Modes in Concrete Joints Using Machine Learning
Author Information
Author(s): Karampinis Ioannis, Karabini Martha, Rousakis Theodoros, Iliadis Lazaros, Karabinis Athanasios
Primary Institution: Democritus University of Thrace
Hypothesis
Can analytical equations derived from SHAP values improve the prediction of failure modes in reinforced concrete beam-column joints?
Conclusion
The study successfully derived analytical equations that predict failure modes in concrete joints with an accuracy comparable to machine learning models.
Supporting Evidence
- The derived analytical equations achieved an accuracy of approximately 78%.
- The precision, recall, and F1-score for the failure mode 'JS' were around 80%.
- The model was not biased towards one of the two failure modes, indicating balanced performance.
Takeaway
This study helps engineers predict how concrete joints will fail during earthquakes, making buildings safer.
Methodology
The study used a dataset of 478 experimental results and applied the eXtreme Gradient Boosting algorithm to predict failure modes, achieving an accuracy of approximately 84%.
Potential Biases
The model may be biased if the dataset is not representative of all possible joint configurations.
Limitations
The study's findings are based on a specific dataset and may not generalize to all types of concrete structures.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website