RISS 학술연구정보서비스

검색

인기 검색어

    다국어 입력

    http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.

    변환된 중국어를 복사하여 사용하시면 됩니다.

    예시)
    • 中文 을 입력하시려면 zhongwen을 입력하시고 space를누르시면됩니다.
    • 北京 을 입력하시려면 beijing을 입력하시고 space를 누르시면 됩니다.
    닫기

    AI 기반 디지털 트윈 기술을 활용한 다중운집인파사고 관리체계 개선 방안 : 안전취약계층을 중심으로

    한글로보기

    https://www.riss.kr/link?id=T17376708

    • 저자
    • 발행사항

      광주: 광주대학교 대학원, 2026

    • 학위논문사항

      학위논문(박사) -- 광주대학교 대학원 , 방재안전학과 , 2026. 2

    • 발행연도

      2026

    • 작성언어

      한국어

    • 주제어
    • DDC

      363.34 판사항(23)

    • 발행국(도시)

      광주

    • 기타서명

      Development of an AI-Driven Digital Twin Framework for Enhancing Crowd Disaster Management Systems: Focused on Safety Vulnerable Groups

    • 형태사항

      115p.: 천연색삽화, 도표; 26cm.

    • 일반주기명

      광주대학교 논문은 저작권에 의해 보호받습니다.
      지도교수:송창영
      참고문헌 수록.

    • UCI식별코드

      I804:24003-200000958568

    • 소장기관
      • 광주대학교 도서관 소장기관정보
    • 0

      상세조회
    • 0

      다운로드
    서지정보 열기
    • 내보내기
    • 내책장담기
    • 공유하기
    • 오류접수
    인용문이 복사되었습니다.

    부가정보

    다국어 초록 (Multilingual Abstract) kakao i 다국어 번역

    Development of an AI-Driven Digital Twin Framework for Enhancing Crowd Disaster Management Systems: Focused on Safety Vulnerable Groups

    Lee DaeJin
    Advisor : Prof. Song Chang-Yeong, Ph.D.
    Department of Disaster Prevention & Safety,
    Graduate School of Gwangju University


    The Itaewon disaster revealed the fundamental weakness of the modern disaster management system. Although surveillance technology and density-based monitoring systems have been introduced everywhere, there are still fatal loopholes in safely managing large crowds. Existing research has been conducted under the premise of a uniform crowd with the same adult, that is, everyone behaves similarly. This method improves the simplicity of management, but in reality, it does not properly reflect the reality that each individuals characteristics are complex and affect collective dynamics in a high-density environment.

    Most of these studies do not reflect safety vulnerable groups such as the elderly, children, and the disabled. Safety vulnerable groups are exposed to risks due to complex limitations in crowded situations. As it has such various vulnerabilities, it greatly reduces the accuracy of disaster prediction. A disaster management system that does not receive fair protection results in deepening disaster inequality in crisis situations.

    Existing studies have focused on crowd density. Individual diversity has often been put on the back burner. As a result, studies so far have been close to follow-up after a problem occurs. In this paper, a study is needed on how to organize the crowd and how the risks vary depending on these configurations.

    These changes go beyond simple technological innovation and lead to ethical discussions that take a new look at crowd safety from the perspective of distribution justice. Based on this, a preemptive AI-based framework based on the disaster justice principle that values fairness and inclusion is being proposed. The system combines computer vision classification model and digital twin simulation environment to identify vulnerable groups in real time and calculate the Safety Vulnerable Class Weighted Risk Index (V-CRI), thereby evaluating their impact on the overall risk.

    By making the composition of the crowd itself an important variable, this approach allows a transition from the existing post-response and density-centered method to a preemptive system. Those in urgent need of protection before an incident can be protected first.

    Keywords : AI, Digital Twin, Crowd Disaster Management, Safety Vulnerable Group
    번역하기

    Development of an AI-Driven Digital Twin Framework for Enhancing Crowd Disaster Management Systems: Focused on Safety Vulnerable Groups Lee DaeJin Advisor : Prof. Song Chang-Yeong, Ph....

    Development of an AI-Driven Digital Twin Framework for Enhancing Crowd Disaster Management Systems: Focused on Safety Vulnerable Groups

    Lee DaeJin
    Advisor : Prof. Song Chang-Yeong, Ph.D.
    Department of Disaster Prevention & Safety,
    Graduate School of Gwangju University


    The Itaewon disaster revealed the fundamental weakness of the modern disaster management system. Although surveillance technology and density-based monitoring systems have been introduced everywhere, there are still fatal loopholes in safely managing large crowds. Existing research has been conducted under the premise of a uniform crowd with the same adult, that is, everyone behaves similarly. This method improves the simplicity of management, but in reality, it does not properly reflect the reality that each individuals characteristics are complex and affect collective dynamics in a high-density environment.

    Most of these studies do not reflect safety vulnerable groups such as the elderly, children, and the disabled. Safety vulnerable groups are exposed to risks due to complex limitations in crowded situations. As it has such various vulnerabilities, it greatly reduces the accuracy of disaster prediction. A disaster management system that does not receive fair protection results in deepening disaster inequality in crisis situations.

    Existing studies have focused on crowd density. Individual diversity has often been put on the back burner. As a result, studies so far have been close to follow-up after a problem occurs. In this paper, a study is needed on how to organize the crowd and how the risks vary depending on these configurations.

    These changes go beyond simple technological innovation and lead to ethical discussions that take a new look at crowd safety from the perspective of distribution justice. Based on this, a preemptive AI-based framework based on the disaster justice principle that values fairness and inclusion is being proposed. The system combines computer vision classification model and digital twin simulation environment to identify vulnerable groups in real time and calculate the Safety Vulnerable Class Weighted Risk Index (V-CRI), thereby evaluating their impact on the overall risk.

    By making the composition of the crowd itself an important variable, this approach allows a transition from the existing post-response and density-centered method to a preemptive system. Those in urgent need of protection before an incident can be protected first.

    Keywords : AI, Digital Twin, Crowd Disaster Management, Safety Vulnerable Group

    더보기

    목차 (Table of Contents)

    • 제1장 서론 1
    • 제1절 연구의 배경 및 필요성 1
    • 1. 연구의 배경 1
    • 2. 연구의 필요성 2
    • 제2절 연구의 목적 및 방법 4
    • 제1장 서론 1
    • 제1절 연구의 배경 및 필요성 1
    • 1. 연구의 배경 1
    • 2. 연구의 필요성 2
    • 제2절 연구의 목적 및 방법 4
    • 1. 연구의 목적 4
    • 2. 연구의 방법 5
    • 제2장 이론적 배경 및 고찰 7
    • 제1절 다중운집인파사고의 개념 및 사례 7
    • 1. 다중운집인파사고 개념 8
    • 2. 주요 국가별 다중운집인파사고 사례 12
    • 3. 주요 국가별 다중운집인파사고 입법사례 16
    • 제2절 다중운집인파사고 관리체계 20
    • 1. 현행 관리체계 분석 20
    • 2. 현행 관리체계의 구조적 한계 21
    • 제3절 다중운집인파사고 관리를 위한 기술 동향 22
    • 1. 디지털 트윈 기술 개요 22
    • 2. AI를 통한 예측 29
    • 제4절 이론적 고찰 31
    • 1. 군중 동역학과 이질적 군중 모델 31
    • 2. 안전취약계층의 특성 33
    • 3. AI 기반 인파 감지 35
    • 제3장 AI 기반 디지털 트윈 시스템 및 안전취약계층 위험 평가 모델(V-CRI) 설계 37
    • 제1절 시스템 아키텍처 37
    • 1. 아키텍처 정의 37
    • 2. 안전취약계층 데이터 처리 42
    • 제2절 AI 기반 안전취약계층 식별 모델 43
    • 1. 안전취약계층 실시간 탐지 및 분류 43
    • 2. 학습데이터 확보 방안과 전처리 46
    • 3. 폐색 상황에서의 식별 49
    • 제3절 안전취약계층 가중 위험지수(V-CRI) 모델 51
    • 1. 밀도 중심 위험 평가의 한계 51
    • 2. 안전취약계층 가중 위험지수(V-CRI) 51
    • 3. V-CRI 모델 가중치 설정 52
    • 4. 안전취약계층 위험 가중치 53
    • 제4절 소결 56
    • 제4장 시뮬레이션 기반 관리체계 개선 58
    • 제1절 시뮬레이션 설계 58
    • 1. 디지털 트윈 환경 구축 및 에이전트 구현 58
    • 2. 시나리오 구성 61
    • 3. 평가지표 67
    • 제2절 시뮬레이션 결과 비교 분석 68
    • 1. 안전취약계층 결과 분석 69
    • 2. 비선형적 패턴 분석 및 임계점 규명 74
    • 제3절 다중운집인파사고 관리체계 개선 77
    • 1. V-CRI 기반 위험 평가 77
    • 2. 안전취약계층 분리 78
    • 3. 현행 관리체계 개선 방안(AS-IS vs TO-BE) 79
    • 제5장 결론 81
    • 참고문헌 83
    • 부록 90
    더보기

    분석정보

    View

    상세정보조회

    0

    Usage

    원문다운로드

    0

    대출신청

    0

    복사신청

    0

    EDDS신청

    0

    동일 주제 내 활용도 TOP

    더보기

    주제

    연도별 연구동향

    연도별 활용동향

    연관논문

    연구자 네트워크맵

    공동연구자 (7)

    유사연구자 (20) 활용도상위20명

    이 자료와 함께 이용한 RISS 자료

    나만을 위한 추천자료

    해외이동버튼