|
Title |
Machine Learning Model for Ductility-Based Blast Damage Assessment of Reinforced Concrete Columns
|
|
Authors |
김예은(Ye-Eun Kim) ; 김수빈(Subin Kim) ; 이기학(Kihak Lee) ; 신지욱(Jiuk Shin) |
|
DOI |
https://doi.org/10.4334/JKCI.2024.36.6.647 |
|
Keywords |
폭발손상평가; 철근콘크리트 기둥; 유한요소해석; 기계학습 blast resistance performance assessment; LS-DYNA; reinforced concrete column; machine learning |
|
Abstract |
This study developed a rapid blast damage assessment engine based on displacement-based blast performance evaluation methods. The engine uses a Multi-Step Learner model combining two machine learning models to enhance the accuracy of blast damage prediction according to failure types. The input variables for the model include column details, blast loading scenarios, and failure types derived through machine learning models. The output variable is the blast damage grade, based on displacement ductility. To develop the model, a training and validation dataset was generated using finite element analysis models from previous studies. Seven classification learners were trained, and the best-fit model was selected based on its superior prediction performance. The ensemble method showed outstanding performance, with F-score and AUC values 41.05 % and 14.65 % higher, respectively, compared to other classifiers on the test dataset.
|