RC 기둥의 폭발 저항성능 예측을 위한 소수 데이터 세트 기반 기계학습 모델 방법론

초록

This study proposes a machine learning model with a combining method capable of accurately evaluating the blast resistance performance of reinforced concrete (RC) columns using a small dataset of 200 samples. To achieve this, a blast performance evaluation response database was established based on finite element analysis models that consider various column details and blast scale values. Each individual learning model applied seven classification algorithms, and the model demonstrating the highest evaluation metrics was developed and combined. The proposed machine learning model achieved a 65.5 % reduction in data usage compared to an existing model based on 700 samples while improving performance by an average of 14.3 %. These results demonstrate that the proposed method enables highly accurate and rapid evaluations even in data-limited environments.

키워드