Projects and Grants per year
Grants and Contracts Details
Description
Residential energy consumption has been rapidly increasing during the last decades, especially in the
U.S. where 2.6 trillion kilowatt-hours have been consumed during the year 2015 alone. Several research
efforts have been made to reduce this consumption, including demand response and smart residential environments.
However, previous work in this context has mainly been performed by overlooking the complexity
of human nature, and the human actions and perceptions when interacting with such systems. In
fact, recent research has shown that these approaches may actually generate negative attitudes, potentially
resulting in an increase of the energy consumption and even abandonment and avoidance of such systems.
The goal of this proposal is to advance the state of the art in residential energy management by designing
algorithmic solutions that learn and take into account human behaviors and perceptions using a variety
of tools such as optimization theory, statistics, machine learning and graph theory. This revolutionary approach
will allow to reduce the energy consumption of the residential sector and to share renewable energy
resources efficiently, while contributing to the NSF big idea “The Future ofWork at the Human-Technology
Frontier” by providing methods to improve the human-technology partnership.
Intellectual Merit: The proposed research has the potential to transform the way in which energy management
for smart residential environments is studied in the field of computer science, by defining and
integrating previously unexamined human behaviors and perceptions in the algorithms’ design and optimization
techniques. In order to enable fine grained energy monitoring, we propose novel stream-based
appliance recognition algorithms for smart outlets. These algorithms not only learn the appliance consumption
signatures, but also the user willingness to interact with the system by trading off representatives
and informativeness of the collected samples with the learned user behavioral model. Using a smart phone
app to collect user feedback developed by undergraduate students as part of their Capstone projects, we
design new learning algorithms based on regressograms, interpolation and regression to quantify the user
perception of appliances and their dependencies. Using the social-behavioral well-being models that the PI
has proposed in his previous interdisciplinary work, we design energy saving optimization strategies that
take into account the user perception. In addition, we consider smart residential environments equipped
with renewable energy generation and sharing capabilities. We propose optimization algorithms to match
users’ demand and production, by taking into account their degree of involvement in the energy exchange
process and their preferences. Finally, we address the problem of reproducibility of survey results. In
this context, we provide for the first time methodologies to quantify the difference between subject pools
that go beyond the standard statistical sampling. The proposed research is validated through real testbeds
and large scale simulations based on real traces. The real testbeds are made available through the Power
and Energy Institute of Kentucky (PEIK), thanks to collaborations with large regional utilities, such as the
Tennessee Valley Authority (TVA) and LGE-KU.
Broader Impacts: The proposed research has the potential of expanding the frontiers of residential energy
management systems by integrating human behaviors and perceptions in the algorithms design and
analysis. Furthermore, the results can benefit society by reducing the energy consumption of the residential
sector. In addition, this proposal will have significant impact in the education of elementary, high school,
undergraduate and graduate students. Specifically, innovative educational activities supported by this proposal
include classes taught by the PI with real time demos, coding challenges and research experiences at
the PI lab for students of the Math, Science, and Technology Center (MSTC) program of the Paul Laurence
Dunbar High School, Lexington, KY. Furthermore, the PI will be joining the STARS IGNITE program, leading
a cohort of students to the diversity-oriented conference Grace Hopper, and teaching periodic seminars
for hispanic elementary students by leveraging the University of Kentucky Society of Hispanic Professional
Engineers students. This society hosts a weekly seminars called (EFL)2 (Engineering Exploratory for
Latinos from Latinos) that focuse on interactive/informational engineering activities with students from
Cardinal Valley Elementary School (76% Hispanic demographic). The proposal also provides support for
the training of graduate and undergraduate students through exciting research projects, and the produced
results will be integrated in the Cyber-Physical-Human Systems class taught by the PI.
1
Status | Active |
---|---|
Effective start/end date | 3/1/20 → 2/28/25 |
Funding
- National Science Foundation: $568,987.00
Fingerprint
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.
Projects
- 3 Active
-
REU Supplement: Participant Support: Energy Management for Smart Residential Environments through Human-in-the-loop Algorithm Design
Silvestri, S. (PI)
3/1/20 → 2/28/25
Project: Research project