The Smart Thermostat: Using Occupancy Sensors to Save Energy in

Report
Jiakang Lu, Tamim Sookoor, Vijay Srinivasan, Ge Gao, Brian Holben, John
Stankovic, Eric Field, Kamin Whitehouse
SenSys’10
 Abstract
 Motivation
 Challenges
 Introduction
 Design
 Experiment Setup
 Evaluation
 Conclusion
 Abstract
 Motivation
 Challenges
 Introduction
 Design
 Experiment Setup
 Evaluation
 Conclusion
 Heating, ventilation and cooling(HVAC) is the




largest source of residential energy consumption
Using cheap($5 each) and simple sensors(motion ,
door sensors) to automatically sense the occupancy
and sleep patterns in a home
Automatically turn on or off the HVAC system
Achieve 28% energy saving on average, at a cost of
approximately $25 in sensors
Commercially-available baseline approach that use
similar sensors saves only 6.8% energy on average
 HVAC is the single largest contributor to a home’s
energy bills and carbon emission
 Accounting for 43% of residential energy consumption
in the U.S. and 61% in Canada and U.K., which have
colder climates
 Recent studies show that households with
programmable thermostats have higher energy
consumption on average than those with manual
controls
 Users program them incorrectly or disable them
altogether
 Quickly and reliably determine when occupants leave
the home or go to sleep
 Motion sensor are notoriously poor occupancy sensors,
which often turn lights off when a room is still occupied
 When to turn HVAC system on
 Too early or too late, both waste energy
 Abstract
 Motivation
 Challenges
 Introduction
 Design
 Experiment Setup
 Evaluation
 Conclusion
 A 2-stage heat pump and
a third stag electric heater
 Stage 2 is the most efficient
stage, but with longer
response time
 Stage 3 has the fastest
response time but high energy cost
 Stage 1 operates at a lower power level, it is more
effective at maintaining a constant temperature
 Problems
 Occupants leave home shortly after 9AM, but the system
continues heating until 10AM
 Shallow setback( typically 5 degrees Celsius)
 Comfort loss
 The risk of comfort loss causes people to reduce their
use of setback schedules
 Uses motion sensors, door sensors, or card key access
systems to turn the HVAC on and off
 4 out of 8 households actually increase energy usage
by up to 10%
 Leave at 9:30AM  turn off at 10:30AM
 Shallow setback
 Inefficient stage of heating
 Fast reaction algorithm uses a probabilistic model to
process the sensor data to estimate the occupancy
 Preheating
 Deep setback( about 10 degrees Celsius)
 Abstract
 Motivation
 Challenges
 Introduction
 Design
 Experiment Setup
 Evaluation
 Conclusion
 Passive infrared(PIR) motion sensors in rooms
 Magnetic reed switches on entry ways
 Approximately $5 each
 Target preheat time  t
 Arrival time  a
 Tulum dataset[13], which was created by monitoring
the occupants of a home for approximately one month
 a < t  heat with stage 3
 a > t  heat with stage 2
 optimum preheating time 18:06
 Tlast  the time elapsed since the last sensor firing
 It changes to idle state when it’s idle time is over a
threshold
 It changes to active state when it detect an event
 “Sleep” from 10PM to 10AM
 Estimate the probability of 3 states of the home
 Away, Active, Sleep
 HMM transitions to a new state every
5 minutes
 xt is a vector of 3 features of sensor data
 time of day at 4-hour granularity
 total number of sensor firings in time interval dT
 binary features to indicate presence of front door,
bedroom, bathroom, kitchen, and living room sensor
firings in dT
 Training the model
 Abstract
 Motivation
 Challenges
 Introduction
 Design
 Experiment Setup
 Evaluation
 Conclusion
 Empirical data traces from 8 instrumented homes
 Occupant surveys of 41 homes(4 weeks)
 Two public smart home dataset
 Tulum, Kasteren
 One motion sensor in each room
 One door sensor on each entryway to the home, and
some inner doors
 Daily interviews with the residents to clarify
ambiguous or questionable data
 Not perfectly accurate
 Previous studies have used approaches ranging from
self reports to video camera recordings
 None of these schemes for creating ground truth are
expected to be perfect
 Empirical data traces from 8 instrumented homes
 Occupant surveys of 41 homes(4 weeks)
 Two public smart home dataset
 Tulum, Kasteren
 Each individual wrote down their sleep, wake, leave,
and arrive times every day
 Retirees, students, professionals, young professionals,
and families
 Empirical data traces from 8 instrumented homes
 Occupant surveys of 41 homes(4 weeks)
 Two public smart home dataset
 Tulum, Kasteren
 Use only the leave, arrival, and sleep event labels
 Performance can be affected by “outdoor temperature”,
“air leakage”, and “house insulation”
 Evaluate different thermostat algorithms under
different housing conditions and climates
 Weather data are from the local airport weather
station that provides hourly data
 Abstract
 Motivation
 Challenges
 Introduction
 Design
 Experiment Setup
 Evaluation
 Conclusion
 Threshold ↑, slow reaction, treat inactive as active
 Threshold ↓, fast reaction, treat active as inactive
 HMM  88% accuracy
 React5  78% accuracy
 Run 14 days in January and July using the climate in
Charlottesville, VA to evaluate both cooling and
heating.
 Deep setbacks to 10°C for heating and 40°for cooling
 Reactive thermostat wastes energy due to frequent
reactions
 Reactive saves 2.9kWh(6.8%), misses 60 mins on average
 Smart saves 11.8kWh(27.9%), misses 48 mins on average
 Using data from “home B”
 Fast reaction, Deep setback, Preheating
 Add deep setback slightly increase miss time
 Save 34% energy, improve miss time by 51 mins
 Almost negligible
 Selected vs. All  energy 28.9% vs. 23.6%
 miss time 54 mins v.s 48 mins
 Selected set performs better!!!???
 Periodic vs. Aperiodic life style
 Save more energy and miss less time for periodic life
style
 As climate becomes warmer from MN to TX, it approaches
the optimal scheme
 Deep setbacks are beneficial during the day warm
region, peak loads at mid-day  cold region, night
 Cold region consumes much more energy
 Lowering same amount of set point 5-8 degrees will only
reduce the energy by a fraction
 Smart thermostat that senses occupancy statistics in a
home to save energy through improved control of the
HVAC system
 Very low cost  less than $25 per home
 save 28% HVAC energy
 Although it just saves $15 per month for a family, it has
nationwide savings about $15 billion annually, and
prevent 1.12 billion tons of pollutants from being
released into the air.
Q&A
Thanks~

similar documents