서브메뉴
검색
Digital Twins at Scale: Accelerating Building Energy Retrofits with Public Data and Machine Learning
Digital Twins at Scale: Accelerating Building Energy Retrofits with Public Data and Machine Learning
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202105618
- ISBN
- 9798265427182
- DDC
- 333.79
- 저자명
- Mayer, Kevin.
- 서명/저자
- Digital Twins at Scale: Accelerating Building Energy Retrofits with Public Data and Machine Learning
- 발행사항
- [Sl] : Stanford University, 2025
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2025
- 형태사항
- 120 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-05, Section: B.
- 주기사항
- Advisor: Fischer, Martin.
- 학위논문주기
- Thesis (Ph.D.)--Stanford University, 2025.
- 초록/해제
- 요약Matching buildings with appropriate retrofit options to make them more energy efficient requires building-specific data. However, this data is often unavailable, preventing retrofits from being deployed at scale and limiting coordination among homeowners, grid operators, and energy auditors. Building on the Digital Twin Framework, this thesis addresses the building-level data gap through two main approaches: (1) extracting relevant building information from widely available public data sources using machine learning and (2) generating plausible building-level data where none exists. These approaches are validated in three case studies. The first presents a method to create a scalable rooftop solar registry using aerial imagery and 3D building data, extending prior work by estimating solar system orientation and tilt. Validated against Germany's official registry, it significantly reduces errors, especially for residential systems (3.6--33.1%) and for systems tilted beyond 40° (25.6--38.1%). The second case study integrates street view, aerial, and thermal imagery with footprint data to classify building energy efficiency. Validated on over 11,900 buildings in the UK and Denmark, it achieves a macro F1 score of 64.64%, outperforming existing heuristics by 9.78%. The third introduces a novel dataset of 6.2 million semantically enriched 3D building models at Level of Detail 4 paired with AI-generated floor plans and roof point clouds, facilitating the development of generative ML algorithms for automated 3D model creation in energy simulations. Together, these contributions show how public data and machine learning can make retrofits more efficient, scalable, and data-driven. Future work is required to refine the presented methods and to empower even more key stakeholders in making sound retrofit decisions at scale.
- 일반주제명
- Energy consumption
- 일반주제명
- Decision making
- 기타저자
- Stanford University.
- 기본자료저록
- Dissertations Abstracts International. 87-05B.
- 전자적 위치 및 접속
- 로그인 후 원문을 볼 수 있습니다.
MARC
008260126s2025 us c eng d■001000017360778
■00520260202105618
■006m o d
■007cr#unu||||||||
■020 ▼a9798265427182
■035 ▼a(MiAaPQ)AAI32316480
■035 ▼a(MiAaPQ)Stanfordwh040yw5612
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a333.79
■1001 ▼aMayer, Kevin.
■24510▼aDigital Twins at Scale: Accelerating Building Energy Retrofits with Public Data and Machine Learning
■260 ▼a[Sl]▼bStanford University▼c2025
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a120 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-05, Section: B.
■500 ▼aAdvisor: Fischer, Martin.
■5021 ▼aThesis (Ph.D.)--Stanford University, 2025.
■520 ▼aMatching buildings with appropriate retrofit options to make them more energy efficient requires building-specific data. However, this data is often unavailable, preventing retrofits from being deployed at scale and limiting coordination among homeowners, grid operators, and energy auditors. Building on the Digital Twin Framework, this thesis addresses the building-level data gap through two main approaches: (1) extracting relevant building information from widely available public data sources using machine learning and (2) generating plausible building-level data where none exists. These approaches are validated in three case studies. The first presents a method to create a scalable rooftop solar registry using aerial imagery and 3D building data, extending prior work by estimating solar system orientation and tilt. Validated against Germany's official registry, it significantly reduces errors, especially for residential systems (3.6--33.1%) and for systems tilted beyond 40° (25.6--38.1%). The second case study integrates street view, aerial, and thermal imagery with footprint data to classify building energy efficiency. Validated on over 11,900 buildings in the UK and Denmark, it achieves a macro F1 score of 64.64%, outperforming existing heuristics by 9.78%. The third introduces a novel dataset of 6.2 million semantically enriched 3D building models at Level of Detail 4 paired with AI-generated floor plans and roof point clouds, facilitating the development of generative ML algorithms for automated 3D model creation in energy simulations. Together, these contributions show how public data and machine learning can make retrofits more efficient, scalable, and data-driven. Future work is required to refine the presented methods and to empower even more key stakeholders in making sound retrofit decisions at scale.
■590 ▼aSchool code: 0212.
■650 4▼aEnergy consumption
■650 4▼aDecision making
■690 ▼a0800
■71020▼aStanford University.
■7730 ▼tDissertations Abstracts International▼g87-05B.
■790 ▼a0212
■791 ▼aPh.D.
■792 ▼a2025
■793 ▼aEnglish
■85640▼uhttp://www.riss.kr/pdu/ddodLink.do?id=T17360778▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.
