Human-centric demand side management : lifestyles, privacy, and fairness
Abstract/Contents
- Abstract
- Power grids are shifting from fossil fuel generation towards renewables such as solar and wind, driven by decarbonization targets. Because of variability and uncertainty in generating power from renewables, we face new challenges of balancing supply and demand in the power grid. To address these challenges, I investigate demand-side management (DSM) because it can be cheaper to run compared with reserving the traditional backup generation or procuring energy in the real-time market. However, some key issues are overlooked in the current DSM like understanding user behavior, preserving user privacy, and preventing discrimination against certain users such as those less able to carry out DSM or respond to pricing (e.g., time-of-use price). This dissertation primarily focuses on demand-side management in electricity systems and presents scalable frameworks to gain insights from household electricity data, to protect private attributes associated with the electricity data, and to promote fairness in managing demand-side services. I obtain household behavioral insights from residential meter data by introducing a new concept--dynamic energy lifestyles--that characterizes behavioral patterns of household energy use in different temporal spans. I also introduce new metrics and machine learning approaches in the context of energy data analysis, both of which are needed to obtain a meaningful number of energy lifestyles. These lifestyles help us to better understand both stability and change patterns of a household's energy use over time. My approach and results can be used by utility companies or energy service providers for identifying households to install rooftop solar and differentiating households' demand flexibility to promote dynamic pricing based on their lifestyles over seasons. To address privacy issues specific to the energy domain, I build a framework that preserves data quality and protects sensitive information. Privacy is quantified by the correlation between sensitive attributes (e.g., income) and the data I need to use. Taking into account the tradeoff between data privacy and data utility and inspired by generative adversarial networks (GAN), I formulate a data sharing task as a game between a data actuator and an adversarial user who aims to infer the sensitive information, then use minimax optimization to alter the raw data. My results indicate that privacy can be preserved with limited performance loss (5%--12%) on data utility tasks. To tackle the challenge of ensuring fairness in DSM, I investigate a use case: engaging users in demand response programs. In this case, privacy restriction must be relaxed, because fairness cannot be obtained by blindness to the protected attribute (e.g., race, income, etc.). I propose a general form of stochastic optimization that treats different groups similarly via fairness constraints in light of uncertain electricity demand. Moreover, when a limited set of demand reductions are revealed, I cast the stochastic optimization into the multi-armed bandit setting and introduce new methods to solve it with sublinear regrets. Overall, I propose conceptual frameworks and develop new methods, all of which are operationalized with data and have the ability to advance human-centric demand side management. Such an impact can help utility operators to plan and provide new energy services, e.g., using dynamic pricing based on residential energy lifestyles, protecting privacy of smart meter data, and promoting energy equity for adopting distributed energy resources.
Description
Type of resource | text |
---|---|
Form | electronic resource; remote; computer; online resource |
Extent | 1 online resource. |
Place | California |
Place | [Stanford, California] |
Publisher | [Stanford University] |
Copyright date | 2022; ©2022 |
Publication date | 2022; 2022 |
Issuance | monographic |
Language | English |
Creators/Contributors
Author | Chen, Xiao |
---|---|
Degree supervisor | Rajagopal, Ram |
Thesis advisor | Rajagopal, Ram |
Thesis advisor | El Gamal, Abbas A |
Thesis advisor | Fischer, Martin, 1960 July 11- |
Degree committee member | El Gamal, Abbas A |
Degree committee member | Fischer, Martin, 1960 July 11- |
Associated with | Stanford University, Civil & Environmental Engineering Department |
Subjects
Genre | Theses |
---|---|
Genre | Text |
Bibliographic information
Statement of responsibility | Xiao Chen. |
---|---|
Note | Submitted to the Civil & Environmental Engineering Department. |
Thesis | Thesis Ph.D. Stanford University 2022. |
Location | https://purl.stanford.edu/qb116kt5660 |
Access conditions
- Copyright
- © 2022 by Xiao Chen
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...