Hao Zhu March 30, 2015 Power & Energy Systems Group Dept. of Electrical & Computer Engineering University of Illinois, Urbana-Champaign Making Sense of Big Data Part 4 Energy Data Disaggregation
About this module Prof. Hao Zhu (haozhu@) Office hours (for ECE 330) every Tuesday 11-12:30 (ECEB 4056) Week 10: motivation and context, data pre-processing Weeks 11-12: disaggregation methods Two TAs: Max Liu (haoliu6@) and Phuc Huynh (pthuynh2@) TA office hours? 2
Nikola Tesla George Westinghouse Thomas Edison James Clerk Maxwell Source: Creative Commons 3
The electric power grid Wikipedia: Power engineering is a subfield of electrical engineering that deals with the generation, transmission, distribution and utilization of electric power. 211,000 miles of transmission lines 230kV 15,600 power plants 830GW load demand Source: www.theenergylibrary.com 4
If I Only Had a Brain GE 2009 Super Bowl Ad; www.youtube.com 5
Power balance One fundamental operational principle is to continuously balance supply and demand to achieve frequency stability Various generation control and scheduling schemes (from seconds to weeks) Source: http://www.okiden.co.jp/english/r_and_d/
The Smarter Grid Electric utilities have been leaders in using technology Supervisory control and data acquisition (SCADA) systems: monitor and operate the high-voltage transmission systems Source: http://www.imageslides.com/technology/gallery/11604-inside-a-power-grid-control-room-(photos) 7
Smart distribution systems Distribution systems traditionally considered to be very passive, with little real-time data and control How does the power company learn that you've lost power? When you call on the phone. An article in the National Geographic magazine S&C IntelliRupter PulseCloser Distribution automation has been making steady advances for many years, a trend that should accelerate with smart grid funding Elster REX digital meter 8
Smart Meters An electronic device that records electric energy consumption in intervals of an hour or less and communicates at least daily back to the utility Utility-level applications: power outage detection/localization http://blog.opower.com/2014/07/data-algorithm-smart-grid-without-smart-meters/ 9
Consumer-level: smart homes? Customers can examine time-specific energy use, see how they compare within their neighborhood, understand how and why energy use varies over time, and ect. My Energy portal provided by Pacific Gas & Electric (PG&E) Energy Saving! 10
Disaggregated energy data Disaggregation allows us to take a whole building (aggregate) energy signal, and separate it into appliance specific data (i.e., plug or end use data). 11
Why appliance-level feedback? 12
Non-intrusive load modeling Power engineers (including RLE, MIT) have investigated it since 1990s Prior approaches: edge detection, real/reactive power signature analysis, and higher-order harmonics analysis Success requires high-precision metering, mainly used for motor diagnostics Steady-state power consumption of a computer and a bank of incandescent lights 13
Recent growth Number of publications rise in last five years http://blog.oliverparson.co.uk/ 14
Disaggregation options Smart Meter is the lowest-cost & lowest installation effort sensor for consumers 15
Data requirements 16
Ultra-high frequency data A recent approach using electromagnetic interference (EMI) at MHz frequency developed at Uwashington Specific sensors add up the costs in prototype systems http://youtu.be/o-sqo8y8xua 17
Belkin energy disaggregation competition A competition ($25k) on Kaggle from Jul 2 to Oct 30, 2013 EMI-based dataset for appliance use detection and classification 18
Smart meter hardware capabilities 19
Implementation options 20
Commercial solutions Bidgely, (formerly MyEnerSave), CA, USA LoadIQ, NV, USA PlotWatt, NC, USA Verlitics, (formerly Emme), OR, USA HOMEBEAT ENERGY MONITOR EI.X Series Monitor 21
Our focus Minute-second resolution of power consumption data Well supported by the existing smart metering infrastructure Reference Energy Disaggregation Data Set (REDD): contains both householdlevel and circuit-level data from 6 US households, over various durations Learning approaches for non-event based disaggregation 22
References Carrie Armel, K., Gupta, A., Shrimali, G., and Albert, A. Is disaggregation the holy grail of energy efficiency? The case of electricity. Energy Policy 52, (2012), 213 234. Carrie Armel, Energy Disaggregation, Precourt Center, Stanford, 2013 Christoper Laughman, et al. "Power signature analysis." IEEE Power and Energy Magazine, 1.2 (2003): 56-63. Steven Shaw, et al. "Nonintrusive load monitoring and diagnostics in power systems." IEEE Trans. Instrumentation and Measurement, 57.7 (2008): 1445-1454. Sidhant Gupta, et al. "ElectriSense: single-point sensing using EMI for electrical event detection and classification in the home." Proc. 12th ACM Intl Conf. on Ubiquitous computing, 2010. 23