autonomous operation and anomaly detection scheme in home...
TRANSCRIPT
Autonomous Operation and Anomaly Detection Scheme in Home IoT based Energy Management System
Dong-Joo KANG(2018, July 18th)
ITCS 2018 ProgramThe International Workshop on the Internet of Things Cybersecurityand Safety
ITCS 2018
Contents
1) IoT & Energy in Interdependence, … and Decentralization (Democratization) of Energy Systems
2) Autonomous Energy Management by HEMS
3) Anomaly Detection based on Context Understanding
2
ITCS 2018
IoT & Energy in Interdependence, …and Decentralization (Democratization) of Energy Systems
3
ITCS 2018
Energy Issue in Movies
4
Arc-reactor for Ironman Suit PV Generation for A.I.
ITCS 2018
5
Energy IoT: interdependence between energy and IoT
Increasing IoT sensors & data centers Increasing energy demand
Increasing energy assets Energy efficiency improved by IoT
technology
ITCS 2018
6
Energy consumption in IT sectors
Source: https://www.statista.com/statistics/683233/electricity-components-of-the-global-it-sector-by-component/
Collection & Storage of Big Data from IoT Sensors
ITCS 2018
Global IT Players on Energy
7
Google Energy
Apple Energy
Solar City with Tesla
ITCS 2018
8
Data Center Campus
Ice Storage
Hot Thermal Storage and generation
Grid Connection
Synchronous condenser
Data Center Substation
Independent Energy System for Data Center Energy SecurityMicrogrid for Data Center
ITCS 2018
9
Smart grid to Microgrid and Nanogrid (HEMS)
(Source: https://www3.nd.edu/~lemmon/courses/ee30372/) (Source: https://cleantechnica.com/2017/11/13/pittsburgh-becoming-laboratory-microgrid-technology/)
Power System (Smart Grid) Microgrid (BEMS, FEMS) for Commercial & Industrial Consumers
Microgrid (CEMS) for Residential Consumers
HEMShttp://www.nuritelecom.com
ITCS 2018
Home Energy Solution by Google
10
Power Meter Home Energy Management System
[Source: https://wccftech.com/google-nest-future-automated-home/]
ITCS 2018
11
Analog Meter Digital (Smart) Meter
Deployment of Smart Meters to Residential Consumers [1]
• 100% of deployment to high-voltage consumers (industrial & commercial) consumers in most developed countries
• Smart meters being deployed to residential consumers
ITCS 2018
12
Internet, IoT
• Italy, Sweden, Ontario(Canada) already reached 100% of AMI deployment• Major ISGAN countries are targeting 100% of AMI deployment by 2020.
→ New market opportunities expected to be opened from 2020→ New service and business models to be required in Smart Grid→ AMI based Home Platform: Energy Storage System (ESS), Distributed Energy
Resource (DER), Home Automation Network (HAN) or Internet of Things (IoT), Electric Vehicles (EV) & Charging Infrastructure
AMI Platform
Physical Systems
Vision for Future Internet
Deployment of Smart Meters to Residential Consumers [2]
ITCS 2018
13
Smart Metering with Home Automation Network
[Source: Provided by Swedish Nation Expert, Magnus Olofsson in ISGAN]
ITCS 2018
Autonomous Energy Management by HEMS
14
ITCS 2018
Basic Concept of HEMS: energy perspective
15
WashingMachine
Air Conditioner
Demand Response
Home Energy Management System (HEMS)
Smart Appliances
Charging/Discharging Schedules
Energy Storage System
Distributed Renewable Energy Sources
Forecasting
Refrigerator
Decision-Making Variable
ITCS 2018
IoT Sensors for Renewable Energy
16
[Source: https://www.huffingtonpost.com/jeremy-rifkin/internet-of-things_b_8306112.html]
Asset Monitoring & Maintenance
Weather Data Monitoring & Collection
Renewable Output Forecasting
IoT sensors
ITCS 2018
17
IoT Sensors for Smart Home: Demand Management
WashingMachine
Air Conditioner
Refrigerator
Human Action = Manual Operation
Autonomous Management
User Behavior AnalysisContext Analysis
Dynamic Pricing
ITCS 2018
Context Vector Creation [1]
18
User Utility (preference)
Real-time Price
Weather Data from IoT Sensors
Renewable Output
User Status from IoT Sensors
Other Service Options
Fridge Washing Machine TV Aircon Computer Light
Context Vector (Quantifying qualitative and quantitative context) Extension to time-horizon Context Matrix
𝑢𝑢𝑖𝑖𝑡𝑡
𝑠𝑠𝑖𝑖𝑡𝑡
𝑝𝑝𝑖𝑖𝑡𝑡
𝑤𝑤𝑖𝑖𝑡𝑡
𝑟𝑟𝑖𝑖𝑡𝑡
𝑜𝑜𝑖𝑖𝑡𝑡
ITCS 2018
19
Context Vector Creation [2]
𝐶𝐶𝑡𝑡 = (𝑢𝑢1𝑡𝑡 ,𝑢𝑢2𝑡𝑡 , 𝑠𝑠1𝑡𝑡 , 𝑠𝑠2𝑡𝑡 , 𝑝𝑝1𝑡𝑡 ,𝑤𝑤1𝑡𝑡,𝑤𝑤2𝑡𝑡,𝑤𝑤3𝑡𝑡, … , 𝑟𝑟1𝑡𝑡, 𝑟𝑟2𝑡𝑡 , 𝑟𝑟3𝑡𝑡,…, 𝑜𝑜1𝑡𝑡 , 𝑜𝑜2𝑡𝑡 , … )
Assuming a context vector (𝐶𝐶𝑡𝑡) at a specific time t,
Here, - 𝑢𝑢1𝑡𝑡 , 𝑢𝑢2𝑡𝑡 : utility functions of 1st and 2nd users- 𝑠𝑠1𝑡𝑡, 𝑠𝑠2𝑡𝑡 : user status modes of 1st and 2nd users- 𝑝𝑝1𝑡𝑡: real-time pricing applied to HEMS (corresponding user or home owner)- 𝑤𝑤1𝑡𝑡,𝑤𝑤2𝑡𝑡,𝑤𝑤3𝑡𝑡, … : weather data such as temperature, humidity, wind speed, etc.- 𝑟𝑟1𝑡𝑡, 𝑟𝑟2𝑡𝑡, 𝑟𝑟3𝑡𝑡, … : (forecasted or actual) renewable outputs by 1st, 2nd, 3rd renewables- 𝑜𝑜1𝑡𝑡 , 𝑜𝑜2𝑡𝑡 : 1st and 2nd service options on the perspective of value-added services
ITCS 2018
Context Vector Creation [3]
20
𝐶𝐶1
𝐶𝐶2
𝐶𝐶3
𝐶𝐶4
𝐶𝐶21
𝐶𝐶22
𝐶𝐶23
𝐶𝐶24
Context Matrix for 1 day
ITCS 2018
21
Day Time
Nig
ht T
ime
Summer Season TV Washing
Machine Aircon Heater Fridge Computer Light
TV TV TV TV TV Fridge Com TV
Washing Machine TV W.M. Aircon W.M. Fridge Com Fridge
Aircon TV W.M. Aircon Aircon Fridge Com Aircon
Heater TV W.M. Aircon OFF Fridge Com Light
Fridge Fridge Fridge Fridge Fridge Fridge Fridge Fridge
Computer Com Com Com Com Fridge Com Com
Light TV Fridge Light Light Fridge Com Light
Appliance On/Off Priority on Competing Situation• Operation based on historical user behavior and preset input Context matrix for operation: competition between appliances
ITCS 2018
2-Stage HEMS Operation Algorithm
22
Day-ahead Price Refrigerator
Heater
Temperature
Humidity
Internalizing Temporal Constraints → Independent Input Variables
Light
Washer
Home Appliance① On-off② Operation Mode
Contextual Constraints
1st Stage
Demand Response
• On-off mode Appliances→ Operation Pattern decided
at 1st stage• Analog mode Appliance→ Optimization at 2nd stage
Combined Heat & Power
Renewable Energy Output
Battery or Storage
2nd Stage
ITCS 2018
Anomaly Detection using Data Redundancy
23
ITCS 2018
Characteristics of Energy Data
24
General Communication Data (Packet based Communication)
Energy (Electric Power) System Data
ITCS 2018
Energy Data & User Behavior
• Brandon J. Murrill, Edward C. Liu, & Richard M. Thompson: Smart Meter Data: Privacy and Cybersecurity, CRS Report for Congress (Prepared for Members and Committees of Congress), Feb. 3rd, 2012
25
• Home appliances have their own usage pattern depending on home user’s life pattern electric energy consumption data
• User behavior forecast & data integrity verification
ITCS 2018
26
Example of State Estimation with Data Redundancy- Comparison between smart meter & individual appliances
ITCS 2018
Daily Datasets of Individual Data Fields (Loads, Prices)
27
ITCS 2018
28
Example of comparing two different datasets
Load data from an agent Price data from an agent
• Two variables are proportional to each other in normal situation
ITCS 2018
Development of Context Analysis Tool with LabVIEW [1]
29
ITCS 2018
30
Development of Context Analysis Tool with LabVIEW [2]
ITCS 2018
Verification at Community Level
31
Auto-correlation: Time-series
Distribution Power System
BEMS(APT, Building) Microgrid, CEMS ESCO FEMS
HEMS
Smart Appliances
MicrogridLevel
Transmission Power System
Multi-agent Concepts
Correlation btw. Home
Cross-sectional Analysis among Neighbors in Community
Time-series Analysis on Historical Data
Verification by Summation
Energy Consumption of Community= Total Sum of Individual Actors’
Energy Consumptions
• Analysis on relationship (with context matrices) between agents on horizontal and vertical axes
ITCS 2018
Conclusion: HEMS of Context Understanding & ADS
• ADS: Anomaly Detection System Anomaly includes error, fraud, intrusion, etc.
• Cost Minimization & Benefit Maximization (Conventional) Context understanding (Proposed) Economic efficiency is no more the only value at home level
decision making• IoT data will contribute to making more concrete context• Context will include the social incentive of each prosumer • Security or resilience to be considered as competing values with
economic benefit • MAS based modeling & ANT (actor network theory)
32