Many school buildings are vulnerable to earthquakes because they were built before mandatory seismic design was applied. This study uses machine learning to develop an algorithm that rapidly constructs an optimal reinforcement scheme with simple information for non-ductile reinforced concrete school buildings built according to standard design drawings in the 1980s. We utilize a decision tree (DT) model that can conservatively predict the failure type of reinforced concrete columns through machine learning that rapidly determines the failure type of reinforced concrete columns with simple information, and through this, a methodology is developed to construct an optimal reinforcement scheme for the confinement ratio (CR) for ductility enhancement and the stiffness ratio (SR) for stiffness enhancement. By examining the failure types of columns according to changes in confinement ratio and stiffness ratio, we propose a retrofit scheme for school buildings with masonry walls and present the maximum applicable stiffness ratio and the allowable range of stiffness ratio increase for the minimum and maximum values of confinement ratio. This retrofit scheme construction methodology allows for faster construction than existing analysis methods.

1. 서 론
2. 기계학습방법론
    2.1 입력변수 및 출력변수
    2.2 철근콘크리트 기둥 파괴유형 예측모델 결정
3. 기계학습기반 보강전략구축
    3.1 구속비 및 강성비
    3.2 기계학습 기반 최적 보강전략 방법론
4. 학교건축물 보강전략 구축
5. 결 론
/ 감사의 글 /
/ REFERENCES /

• 김수빈(경상국립대학교 건축공학과 박사과정) | Kim Subin (Ph. D. Student, Department of Architecture Engineering, Gyeongsang National University)
• 최인섭(계명대학교 건축공학과 조교수) | Choi Insub (Assistant Professor(PhD), Department of Architectural Engineering, Keimyung University)
• 신지욱(경상국립대학교 건축공학과 부교수(공학박사)) | Shin Jiuk (Associate Professor(PhD), Department of Architecture Engineering, Gyeongsang National University) Corresponding author