EE392N Intelligent Energy Systems: Big Data and Energy April 2, 2013 Dan O’Neill Dimitry Gorinevsky Seminar Course 392N ● Spring2013 ee392n - Spring 2013 Stanford University Intelligent Energy Systems: Big Data O’Neill and Gorinevsky 1
EE392N Intelligent Energy Systems:
Big Data and Energy
April 2, 2013
Dan O’Neill Dimitry Gorinevsky
Seminar Course 392N ● Spring2013
ee392n - Spring 2013 Stanford University
Intelligent Energy Systems: Big Data O’Neill and Gorinevsky
1
Today’s Program
• Class logistics
• Introductory lecture on Intelligent Energy Systems: Big Data and Energy
Intelligent Energy Systems: Big Data O’Neill and Gorinevsky
ee392n - Spring 2013 Stanford University
2
Instructors
• Daniel O’Neill, Consulting Professor in EE
– Network Management and Machine Learning in energy
– Executive and Startup experience
– www.stanford.edu/~dconeill
• Dimitry Gorinevsky, Consulting Professor in EE
– Big Data Analytics for energy and aerospace
– Information Decision and Control Applications in many industries
– www.stanford.edu/~gorin
Intelligent Energy Systems: Big Data O’Neill and Gorinevsky
ee392n - Spring 2013 Stanford University
3
Class Logistics
• 1 unit graded or CR/NC
– Attendance
– Pre-requisites
– Submitting an one page report on in the end
• Weekly on Tuesdays
– The room and time might change!
– Watch the class website announcements
• Introductory lecture - today
• Nine lectures by industry leaders
Intelligent Energy Systems: Big Data O’Neill and Gorinevsky
ee392n - Spring 2013 Stanford University
4
Planned Lectures
April 2, Introductory Lecture, Dan ONeill and Dimitry Gorinevsky, Stanford
April 9, Feature Discovery in Energy Data, Sachin Adlakha, Ayasdi
April 16, The Industrial Internet, Marco Annunziata, GE Energy
April 23, Risk Analytics Applications and Vision, Chris Couper, IBM
April 30, Power Plant Big Data Optimization, Jim Schmid, GE Energy
May 7, Monitoring of T&D Systems, Paul Myrda, EPRI
May 14, Energy Insights from Big Data, Drew Hylbert and Jeff Kolesky, OPower
May 21, AMI Data Management and Analysis, Aaron DeYonker, eMeter/Siemens
May 28, Smart Grid Analytics, Ed Abbo and Houman Behzadi, C3 Energy
June 4, Data Center Energy Management, Manish Marwah, HP
Intelligent Energy Systems: Big Data
O’Neill and Gorinevsky ee392n - Spring 2013 Stanford University
5
Computing and
Communication
Intelligent Energy Systems
ee392n - Spring 2013 Stanford University
Intelligent Energy Systems: Big Data O’Neill and Gorinevsky
6
Energy System
Analytics: Software Function
Energy Systems
• Power systems
• Physical systems in the focus of the class
ee392n - Spring 2013 Stanford University
Intelligent Energy Systems: Big Data O’Neill and Gorinevsky
7
The Traditional Grid
• Worlds Largest Machine! – 3300 utilities
– 15,000 generators, 14,000 TX substations
– 211,000 mi of HV lines (>230kV)
– SCADA control
– Mostly unidirectional
• Capacity constrained graph
Intelligent Energy Systems: Big Data O’Neill and Gorinevsky
ee392n - Spring 2013 Stanford University
8
Interconnect
Intelligent Energy Systems: Big Data O’Neill and Gorinevsky
ee392n - Spring 2013 Stanford University
9
• Renewables
• Demand Response
• Power Flow Management
Nearer Term Initiatives
Intelligent Energy Systems: Big Data O’Neill and Gorinevsky
ee392n - Spring 2013 Stanford University
10
Renewables: The System Problem
National Renewable Energy Laboratory
Intelligent Energy Systems: Big Data O’Neill and Gorinevsky
ee392n - Spring 2013 Stanford University
11
Demand Response
Campus and Buildings Home
• AMI – Advanced Metering Infrastructure • EMS – Energy Management System • Smart devices
Intelligent Energy Systems: Big Data O’Neill and Gorinevsky
ee392n - Spring 2013 Stanford University
12
Power Flow Management
• Adjusting supply
• Routing power flow
• Managing demand
– for aggregated users
– for commercial buildings
Intelligent Energy Systems: Big Data O’Neill and Gorinevsky
ee392n - Spring 2013 Stanford University
13
Computing and Communications
14
Conventional Electric Grid
Generation
Transmission
Distribution
Load
Source IPS
energy subnet
Intelligent Power
Switch
IES
IES
IES IES
Intelligent Energy Network
Conventional Internet
ee392n - Spring 2013 Stanford University
Intelligent Energy Systems: Big Data O’Neill and Gorinevsky
Power and Data Flow
Generators
Transmission 275-400’s KV
Industrial Commercial Business Residential
Distribution 10-20KV to 120V
ISO
Substations
Fiber and uWave
Promise of AMI
Limited visibility
Intelligent Energy Systems: Big Data O’Neill and Gorinevsky
ee392n - Spring 2013 Stanford University
15
Today…
• Integrated platforms
– IBM, Oracle. …
– GE, Seimens, SAIC,…
– Itron, Cisco, …
• But
– Analytic tools
– Vertical analytic products
– Data integration
ee392n - Spring 2013 Stanford University
Intelligent Energy Systems: Big Data O’Neill and Gorinevsky
16
Visualization
Analytics
Elastic Computing
Data
Comm. Platform
Networking
Business Logic
Analytics
ee392n - Spring 2013
Stanford University
Intelligent Energy Systems: Big Data
O’Neill and Gorinevsky 17
Database
Presentation Layer
Computer
Tablet Smart
phone
Internet Communications
Energy Application
Analytics
(Intelligent Functions)
ee392n - Spring 2013
Stanford University
Intelligent Energy Systems: Big Data
O’Neill and Gorinevsky 18
Feedback Control Functions
• Closed loop update
Physical system
Measurement System Sensors
Control Logic
Control Handles Actuators
Plant
ee392n - Spring 2013
Stanford University
Intelligent Energy Systems: Big Data
O’Neill and Gorinevsky 21
Monitoring and Decision Support Functions
Physical system
Monitoring & Decision
Support
Physical
system
Measurement System, Sensors
Data Presentation
• Open-loop functions - Results are presented to an operator
Decision Support Applications
Supply
• Asset Management
• Power Quality Monitoring
• Outage Management
• Risk Management
• Renewables Integration – forecasting
Demand
• Energy Efficiency Monitoring
• Revenue Protection
ee392n - Spring 2013 Stanford University
22 Intelligent Energy Systems: Big Data
O’Neill and Gorinevsky
Plant/System Data
Monitoring Approaches
• Most used approach is Exceedance Monitoring
• Example: grid frequency deviation from 60Hz
ee392n - Spring 2013
Stanford University
Intelligent Energy Systems: Big Data
O’Neill and Gorinevsky 23
Monitoring Function: Exceedance
Detected Anomalies
Big Data
• Tens of Tb to Pb range
• Sequential or parallel processing of data chunks. Each chunk fits into memory
• Much of earlier work involved ‘soft’ data
• Energy: Machine-to-Machine (M2M) data
• Monitoring and decision support applications
• Data mining
ee392n - Spring 2013 Stanford University
Intelligent Energy Systems: Big Data O’Neill and Gorinevsky
24
Data Mining Functions
• Data Exploration
– Performed interactively
• Model Training
– Known as system identification in control
• Model Exploitation
– Estimation, eg, forecasting
– Decision support, eg, monitoring
– Control, eg, embedded optimization
ee392n - Spring 2013 Stanford University
Intelligent Energy Systems: Big Data O’Neill and Gorinevsky
25