Download presentation
1 -

Recent Advances in Algorithms for Energy DisaggregationJ ZicoKolterMIT


J ZicoKolterRecent Advances in Algorithms for Energy DisaggregationThe GoalDetermine breakdown of power given wholehome consumption eg from smart meterTotal PowerJ ZicoKolterRecent Advances in Algorit

beatrice's Recent Documents

Providers
Providers

Children146s Hospital Los Angeles is ranked among the top hospitals in the nation for endocrinology and diabetes care on the US News World Report Best Children146s Hospitals list As one of the larges

published 0K
Executive Summary
Executive Summary

2urpose and Overviewection I Backgroundection II Evidence and Rationale for Evaluation Outcome Criteria UpdatesACurrent Evaluation CriteriaBProcess for Updating Evaluation CriteriaCUpdates Evaluation

published 0K
NAMEDEPARTMENT PONE x000b203x000cAObanox000f Enauman ServLces 5x001c4
NAMEDEPARTMENT PONE x000b203x000cAObanox000f Enauman ServLces 5x001c4

Treasurer 24x001a-6x001c12BucNnerx000f BenParNs 5x001c4-3100BuLOdLnJ Dept BuLOdLnJ5x001c4-3012CaOdareOOax000f SebastLanAssessor5x001c4-3008aLserx000f CKrLsT - nformatLon TecKnoOoJ 5x001c4-302x001aap

published 0K
HeinOnline  93 Harv L Rev  1510 19791980
HeinOnline 93 Harv L Rev 1510 19791980

HeinOnline -- 93 Harv L Rev 1511 1979-1980HeinOnline -- 93 Harv L Rev 1512 1979-1980HeinOnline -- 93 Harv L Rev 1513 1979-1980HeinOnline -- 93 Harv L Rev 1514 1979-1980HeinOnline -- 93 Harv L Rev

published 0K
x0000x00001 xMCIxD 0 xMCIxD 0 Higher Education Emergency Relief Fund
x0000x00001 xMCIxD 0 xMCIxD 0 Higher Education Emergency Relief Fund

Provided supplemental fundingFor existingrivateonprofit and ublic CARES Act grantees theA authorizedsupplemental awardsto existing Student Aid Portion and Institutional Portion grantheseinstitutions w

published 0K
FOREWORDvACKNOWLEDGMENTS viiABOUT THE AUTHORS ixSUMMARYINTRODUCTION
FOREWORDvACKNOWLEDGMENTS viiABOUT THE AUTHORS ixSUMMARYINTRODUCTION

internal and regional conict become so widespread and so brutal in the region What would more accountable social contracts between citizens and states look like and how can Arab countries take advanta

published 0K
School Improvement Plan
School Improvement Plan

Page130Guidelines and ProcessISchool Level NarrativeSchool Building InformationLocal Education Agency LEANameCheltenham School DistrictSchool Building NameCheltenham High School4-Digit School Building

published 0K
ATRICK OF THE
ATRICK OF THE

ST PISLANDWEEKSOF SEPTEMBER 192021PARISH BULLETIN25THSUNDAY IN ORDINARY TIMEMASSESSATURDAY SEPTEMBER 1820215 pm Holy Souls from MariaBlessingof the Wilson FamilySUNDAY SEPTEMBER 1920219amVal Hulse fro

published 0K
Download Section

Download - The PPT/PDF document "" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.






Document on Subject : "Recent Advances in Algorithms for Energy DisaggregationJ ZicoKolterMIT"— Transcript:

1 Recent Advances in Algorithms for Energy
Recent Advances in Algorithms for Energy DisaggregationJ. ZicoKolterMIT Computer Science and Artificial Intelligence LaboratoryBECC Conference, 2011In collaboration with: TommiJaakkola, Matthew J. Johnson,

2 Andrew Y. Ng, SidharthaBatra J. ZicoKolt
Andrew Y. Ng, SidharthaBatra J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation The GoalDetermine breakdown of power given wholehome consumption (e.g. from smart meter) Total Power J. Zico

3 KolterRecent Advances in Algorithms for
KolterRecent Advances in Algorithms for Energy Disaggregation The GoalDetermine breakdown of power given wholehome consumption (e.g. from smart meter) Total Power Disaggregated Power J. ZicoKolterRecent Adv

4 ances in Algorithms for Energy Disaggreg
ances in Algorithms for Energy Disaggregation “EventBased” Detection J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation “EventBased” Detection J. ZicoKolterRecent Advan

5 ces in Algorithms for Energy Disaggregat
ces in Algorithms for Energy Disaggregation “EventBased” Detection Dryer? Fridge? Microwave? Noise? … J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation “EventBased

6 8; Detection Dryer? Fridge? Microwave? N
8; Detection Dryer? Fridge? Microwave? Noise? … J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation “EventBased” Detection Dryer? Fridge? Microwave? Noise? … The standar

7 d paradigm for energy disaggregation e.g
d paradigm for energy disaggregation e.g, Hart, 1992; Leeb, 1993; Laughmanet al., 2003; Patel et al., 2006, Bergéset al., 2010; Gupta et al., Problems: doesn’t provide actual energy breakdown, difficu

8 ltly handling longterm devices, can make
ltly handling longterm devices, can make physically impossible predictions (e.g., device comes on twice without turning off) J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation “Power

9 Packing” Detection J. ZicoKolterRec
Packing” Detection J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation “Power Packing” Detection J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation “Po

10 wer Packing” Detection Coffee Machi
wer Packing” Detection Coffee Machine Dryer Computer J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation “Power Packing” DetectionHelped by nature of AC Power CurrentTime Cof

11 fee Machine Dryer Computer J. ZicoKolter
fee Machine Dryer Computer J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation “Power Packing” Detection J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation “

12 ;Power Packing” Detection A relativ
;Power Packing” Detection A relatively recent approach (Suzuki et al., 2008; Liang et al., 2010; Inagaki et al., 2011)Provides true energy breakdown, can handle “always on” devicesProblems: c

13 ombinatorial (hard) optimization problem
ombinatorial (hard) optimization problem, extremely sensitive to noise/transients and devices that are too similar Our work: Algorithms for “Joint” Energy DisaggregationnKolteret al., 2010; Kolter

14 and Jaakkola, 2011] J. ZicoKolterRecent
and Jaakkola, 2011] J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation Interconnected Prediction Tasks J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation Interconnected P

15 rediction Tasks J. ZicoKolterRecent Adva
rediction Tasks J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation Modeling Interconnected TasksPowerful methods for handling such tasks: probabilistic graphical modelsExactly solving resu

16 lting optimization problem still intract
lting optimization problem still intractable Factorial Hidden Markov ModellGhahramaniand Jordan, 1997] J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation Modeling Interconnected TasksPower

17 ful methods for handling such tasks: pro
ful methods for handling such tasks: probabilistic graphical modelsExactly solving resulting optimization problem still intractable Factorial Hidden Markov ModellGhahramaniand Jordan, 1997] Our method prese

18 nts an approach for approximately solvin
nts an approach for approximately solving such optimization problems; based on methods from convex optimization, probabilistic inferenceIncorporates both the “eventbased” features and the “po

19 wer packing” features, classifies a
wer packing” features, classifies all energy jointlyTractable for large numbers of devices, many states for each device, and long time periods J. ZicoKolterRecent Advances in Algorithms for Energy Disa

20 ggregation Performance Total Power J. Zi
ggregation Performance Total Power J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation PerformanceTrue Disaggregation J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation Pe

21 rformanceOur Method J. ZicoKolterRecent
rformanceOur Method J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation Performance J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation Performance Over Two Weeks Device (C

22 ircuit)Our MethodPrevious ApproachMicrow
ircuit)Our MethodPrevious ApproachMicrowave97.5%94.8%Bath GFI15.0%BathGFI95.1%Bath GFI94.7%37.4%Electronics41.6%39.4%Kitchen Outlets37.5%9.7%Furnace91.7%18.5%KitchenOutlets45.2%12.5%Washer / Dryer98.8%89.3%

23 Total J. ZicoKolterRecent Advances in Al
Total J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation Performance Over Two Weeks Device (Circuit)Our MethodPrevious ApproachMicrowave97.5%94.8%Bath GFI15.0%BathGFI95.1%Bath GFI94.7%37.4

24 %Electronics41.6%39.4%Kitchen Outlets37.
%Electronics41.6%39.4%Kitchen Outlets37.5%9.7%Furnace91.7%18.5%KitchenOutlets45.2%12.5%Washer / Dryer98.8%89.3%Total Publicly Available Data and the Behavior / Algorithms Interplay J. ZicoKolterRecent Advan

25 ces in Algorithms for Energy Disaggregat
ces in Algorithms for Energy Disaggregation REDD: A Publicly Available Dataset eference nergy isaggregation ata Set 16.6 kHz wholehome current and voltagez power consumption for each circuit (Powerhouse D

26 ynamics eMonitor1 Hz power consumption o
ynamics eMonitor1 Hz power consumption on select outlets (EnmetricPowerPortInitial elease (6 homes, ~2 weeks) available at:http://redd.csail.mit.edu Kolterand Johnson, 2011] J. ZicoKolterRecent Advances

27 in Algorithms for Energy Disaggregation
in Algorithms for Energy Disaggregation REDD: A Publicly Available Dataset eference nergy isaggregation ata Set 16.6 kHz wholehome current and voltagez power consumption for each circuit (Powerhouse Dynam

28 ics eMonitor1 Hz power consumption on se
ics eMonitor1 Hz power consumption on select outlets (EnmetricPowerPortInitial elease (6 homes, ~2 weeks) available at:http://redd.csail.mit.edu �12 GB of data, per home, per dayCompress using gre

29 edy approximation to total variation reg
edy approximation to total variation regularization Kolterand Johnson, 2011] J. ZicoKolterRecent Advances in Algorithms for Energy Disaggregation Behavior and Algorithms Behavioral Studies Algorithmic Advan