ERI: Data-Driven Analysis and Dynamic Modeling of Residential Power Demand Behavior: Using Long-Term Real-World Data from Rural Electric Systems

作为分布式能源（例如，屋顶太阳能、能源储存）在家庭中继续增长，更多的房屋将充当虚拟发电厂，将电力送回电网，并显著改变电力系统的运作。现有的负荷分析和预测模型未能充分反映用户的特性，这是至关重要的了解居民电力需求模式。为了提高电力系统的可靠性和稳定性，该项目引入了一种新的数据驱动方法，使用真实世界的长期数据来分析和明确建模家庭层面的电力需求行为。这一研究在理论和实践上都具有重要意义。首先，所提出的模型可以很容易地扩展到其他领域，如用水。其次，它通过调和关于改变消费者行为，特别是绿色消费的激励有效性的混合结果，对营销文献做出了重大贡献。第三，将开发新的“智能电网与数据科学”课程，培养学生将电力工程与社会科学相结合的兴趣。每年将为K-12学生组织数据驱动建模比赛，以鼓励更多学生，特别是女性，学习电力工程和数据科学。将开发一个联合电力工程项目，以促进STEM研究和教育，增加美国土著在STEM职业中的代表性。该项目可以帮助电力合作社开发基于行为的需求响应计划，减少高峰需求费用，并为全球脱碳目标做出贡献。最后，该项目支持美国电力公司制定有效的激励计划，确保公司利润和消费者的可持续用电。研究目标是开发一种新的数据驱动方法，通过将15分钟分辨率的智能电表数据与消费者调查相匹配，沿着当地包裹数据和气象数据，以实现在分解水平上对居民电力需求行为的全面理解。这将为下一代电网的数据驱动分析和建模研究奠定基础。本研究的主要工具是利用奇异值分解（SVD）来辨识和描述电力需求行为。提出了一种基于奇异值分解的特征需求行为提取方法，利用15分钟智能电表数据进行特征需求行为提取。该项目的主要贡献将是开发一个数据驱动的基于温度-时间-日（TTD）的住宅电力需求行为模型，仅依赖于真实世界的数据。电力需求行为将被表示为一个离散时间动态系统，室外温度和一天中的时间坐标随天而变化。该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

As distributed energy resources (e.g., rooftop solar, energy storage) continue to grow among households, more houses will act as virtual power plants, sending electricity back to the power grid and significantly changing power system operations. Existing load analysis and forecasting models fail to adequately reflect consumer characteristics, which are crucial for understanding residential power demand patterns. To improve power system reliability and stability, this project introduces a novel data-driven approach to analyze and explicitly model power demand behavior at the household level using real-world long-term data. This research has several significant impacts on both theory and practice. First, the proposed model can be readily expanded to other domains, such as water usage. Second, it significantly contributes to the marketing literature by reconciling mixed findings on incentive effectiveness in changing consumer behavior, particularly in green consumption. Third, a new "Smart Grid and Data Science" curriculum will be developed, fostering student interest in combining power engineering and social science. Annual data-driven modeling competitions for K-12 students will be organized to encourage more students, especially females, to study power engineering and data science. A joint power engineering program will be developed to promote STEM research and education, increasing Native American representation in STEM careers. This project can help electric co-ops develop behavior-based demand response programs, reducing peak demand charges and contributing to global decarbonization goals. Finally, the project supports U.S. electric utilities in creating effective incentive programs that ensure both firm profit and consumers’ sustainable use of electricity.The research goal is to develop a new data-driven approach to reveal macro- and micro-residential power demand patterns, through matching 15-minute resolution smart meter data with consumer surveys, along with local parcel data and meteorological data, to achieve a comprehensive understanding of residential power demand behavior at a disaggregated level. This will lay the foundation for research focused on data-driven analysis and modeling for next-generation power grids. The key tool employed in the project is utilizing the Singular-Value-Decomposition (SVD) to identify and characterize power demand behavior. An Eigen Demand Behavior Extraction method, based on SVD, is proposed for characterizing demand behavior using 15-minute smart meter data. The primary contribution of this project will be the development of a data-driven Temperature-Time-Day-based (TTD) model for residential power demand behavior, relying solely on real-world data. The power demand behavior will be represented as a discrete-time dynamical system with outdoor temperature and time of a day coordinates evolving over days.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.