서브메뉴
검색
Forecasting at Scale with Human Interactions: Interpretable Artificial Intelligence and an Applied Solution to Scaling Electricity Demand Prediction at Transmission Systems Operators
Forecasting at Scale with Human Interactions: Interpretable Artificial Intelligence and an Applied Solution to Scaling Electricity Demand Prediction at Transmission Systems Operators
상세정보
- 자료유형
- 학위논문 서양
- 최종처리일시
- 20260202105626
- ISBN
- 9798265428028
- DDC
- 500
- 저자명
- Triebe, Oskar.
- 서명/저자
- Forecasting at Scale with Human Interactions: Interpretable Artificial Intelligence and an Applied Solution to Scaling Electricity Demand Prediction at Transmission Systems Operators
- 발행사항
- [Sl] : Stanford University, 2025
- 발행사항
- Ann Arbor : ProQuest Dissertations & Theses, 2025
- 형태사항
- 258 p
- 주기사항
- Source: Dissertations Abstracts International, Volume: 87-05, Section: B.
- 주기사항
- Advisor: Rajagopal, Ram.
- 학위논문주기
- Thesis (Ph.D.)--Stanford University, 2025.
- 초록/해제
- 요약Transmission system operators (TSOs) depend on accurate electricity demand forecasts to ensure the reliable operation of the power grid. The growth of renewable and distributed energy resources is increasing load volatility and localized infrastructure risks, necessitating the scaling of forecasts from a handful of zones to tens of thousands of nodes. Control room operators skillfully assess risks and adjust forecasts; however, human capacity is inherently limited, making it challenging to scale the volume of load forecasts without sacrificing oversight. High-stakes operational decisions also require a comprehensive understanding of forecasts and their underlying model, yet many forecasting methods trade interpretability for accuracy, particularly in large-scale systems.This dissertation advances scalable and interpretable time series forecasting and the deployment of such methods in TSO operations. It introduces a model-building framework that integrates intuitive time series structure with flexible machine learning, employs GPU-accelerated joint training, and supports multi-horizon and multi-series forecasting. The resulting modular hybrid models improve forecast accuracy without compromising interpretability. The framework is embedded within a human-centered workflow and released as an open‑source package.In collaboration with a TSO, an operator-centric, multi-level forecasting system is developed for the zonal-to-nodal transition and validated against extensive real-world operations data. The system enhances accuracy, remains operationally manageable and interpretable at scale, and enables precise monitoring, adjustment, and diagnosis of forecasts. Organized by load behavior in addition to geography, aligning with operators' mental models, the system yields operator-validated, actionable insights.
- 일반주제명
- Decomposition
- 일반주제명
- Seasonal variations
- 일반주제명
- Forecasting
- 일반주제명
- Neural networks
- 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■001000017360837
■00520260202105626
■006m o d
■007cr#unu||||||||
■020 ▼a9798265428028
■035 ▼a(MiAaPQ)AAI32316557
■035 ▼a(MiAaPQ)Stanfordzt934vb5032
■040 ▼aMiAaPQ▼cMiAaPQ
■0820 ▼a500
■1001 ▼aTriebe, Oskar.
■24510▼aForecasting at Scale with Human Interactions: Interpretable Artificial Intelligence and an Applied Solution to Scaling Electricity Demand Prediction at Transmission Systems Operators
■260 ▼a[Sl]▼bStanford University▼c2025
■260 1▼aAnn Arbor▼bProQuest Dissertations & Theses▼c2025
■300 ▼a258 p
■500 ▼aSource: Dissertations Abstracts International, Volume: 87-05, Section: B.
■500 ▼aAdvisor: Rajagopal, Ram.
■5021 ▼aThesis (Ph.D.)--Stanford University, 2025.
■520 ▼aTransmission system operators (TSOs) depend on accurate electricity demand forecasts to ensure the reliable operation of the power grid. The growth of renewable and distributed energy resources is increasing load volatility and localized infrastructure risks, necessitating the scaling of forecasts from a handful of zones to tens of thousands of nodes. Control room operators skillfully assess risks and adjust forecasts; however, human capacity is inherently limited, making it challenging to scale the volume of load forecasts without sacrificing oversight. High-stakes operational decisions also require a comprehensive understanding of forecasts and their underlying model, yet many forecasting methods trade interpretability for accuracy, particularly in large-scale systems.This dissertation advances scalable and interpretable time series forecasting and the deployment of such methods in TSO operations. It introduces a model-building framework that integrates intuitive time series structure with flexible machine learning, employs GPU-accelerated joint training, and supports multi-horizon and multi-series forecasting. The resulting modular hybrid models improve forecast accuracy without compromising interpretability. The framework is embedded within a human-centered workflow and released as an open‑source package.In collaboration with a TSO, an operator-centric, multi-level forecasting system is developed for the zonal-to-nodal transition and validated against extensive real-world operations data. The system enhances accuracy, remains operationally manageable and interpretable at scale, and enables precise monitoring, adjustment, and diagnosis of forecasts. Organized by load behavior in addition to geography, aligning with operators' mental models, the system yields operator-validated, actionable insights.
■590 ▼aSchool code: 0212.
■650 4▼aDecomposition
■650 4▼aSeasonal variations
■650 4▼aForecasting
■650 4▼aNeural networks
■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=T17360837▼nKERIS▼z이 자료의 원문은 한국교육학술정보원에서 제공합니다.
