The views expressed in this presentation are the views of the author and do not necessarily reflect the
views or policies of the Asian Development Bank Institute (ADBI), the Asian
Development Bank (ADB), its Board of Directors, or the governments they represent. ADBI does not guarantee the accuracy of the data included in this paper and accepts no responsibility
for any consequences of their use. Terminology used may not necessarily be consistent with ADB official terms.
Energy Data × Machine Learning
Beyond Energy
Using energy data analytics
to create a smart IoT Ecosystem
About Us
NIALM
World’s First Robot Pet goes independent
AIBO with Artificial NIALM technology with continuous partnership with
Intelligence Evolved from AIBO
Established: April 2013
1999 2009 2010 2011 2013 CEO: Taro Tadano
Office: Tokyo, Japan
Main Investor: Japanese PE fund
Staff: 20 (mainly ex-Sony engineers)
starts Participated in
Energy business Director: Taro Fukumoto, Takashi Kinoshita, Josh Honda
R&D Project Senior Board Advisor: Jay Chinnadorai
Austin Texas
2
Home Energy ICT ECO System
is not working
Customer View Industry, Gov, View
SMART METER
value ＜ Cost
Yet,
3
40% of Electricity consumed in Residential Sector
“NILM” : The Key Technology for Smarter Grid
Non-Intrusive Wide interest in World Wide
「NILM（Non-Intrusive Load Monitoring）」
Measure Load/Appliance
main only …
* And a lot of Ventures..
Result = Itemized Power per Appliance
Total Consumption
Power Element
Disaggregation
Creating a new Home IoT Network
using just a single Smart Sensor and AI in the Cloud
Works on ‘legacy’
appliances without the
need of a Smart Plug
Multi-platform reporting:
Online, Mobile, Tablet
May the “House”
be with you!
3 key ingredients by our Solution
Simple & Scalable
Single Clamp or Embedded into Smart Meter
Evolvable Data Interpretation
Waveform characteristic analysis in Cloud
Wider & Smart usage of data
Usage of data beyond energy consumption
6
contribute to
Sustainable Energy ICT penetration
環境・エネルギーのためのビッグデータフル活用
via
Energy節電の経済効果と人々のエコ意識だけでは自立普及が困難
Data Value Expansion beyond Energy
エネルギーインフラとデータが創り出す新しい自立エコシステムへ
Sustainable
Energy
Infrastructure
BIG DATA
Appliance Diagnosis, Security Support,
Research & Marketing
How does our solution work?
Using a single Smart Sensor inside the distribution board, we can estimate "what appliances", "when", "how
much" power is being consumed at home. Our solution also “Auto Labels” major appliances
Palm-sized (60g)
We can also embed our technology inside a Smart Meter – eliminating the need for any additional hardware!
8
How Our Disaggregation works
Realtime
Hi Resolution
Sampling
First wave form data
of every second
(kHz order sampling)
1 second of data
50hz = 50 waves
Appliance level
disaggregation shows
Electricity wave form data ON/Off status and its
from a single clamp sensor operation mode
9
“High Resolution Analysis” Adds value to both
Utility and Consumer
１．Enabling a Smart Grid using real time and Detail Data
(1) Detailed Load Analysis inc. Power Factor based on waveform data
(2) Highlighting ’unclear’ power consumption to consumers
(3) Enabling Demand Response and behaviour induced Peak Shift
(4) Sub-station* level monitoring and demand forecast
２．Potential Consumer Value Added services
(1) Diagnostics of Home Appliances
① Notification of maintenance request or imminent failure
② “Eco test” and recommendation for replacement
③ Brand/Model level identification – for marketing and/or recall purposes
④ Warning of potential fire hazards (eg. Hob left ON for too long)
⑤ Power system anomaly like arc tracking phenomenon (eg. Disaster alert, Insurance)
(2) Deriving value from energy based Lifestyle Data
① Non-intrusive monitoring of children and the elderly
② Inferring demographic data of household based on ‘appliance activity’
Enabling the provision of marketing, insurance and health care data services
* A challenge for Utilities when PVs are widespread 5
Embedding our solution inside a Smart Meter
By ‘embedding’ our technology into the firmware of a standard smart meter – reduces h/w cost
Waveform data
measurement function
‘embedded’ inside a
standard Smart Meter
WITHOUT
MAJOR COST
INCREASE!!
This Prototype was developed with funding from NEDO and utilises a standard Smart Meter with it’s firmware updated by Informetis
11
Informetis Sensors as a “Home Agent”
Informetis sensors work as a “Home Agent,” a sensory nervous system which contributes to
communication allowing a more natural lifestyle.
12
Our standard application : Magic Home View
A smart Agent for your Home
Integration with Device Control
(Optional)
13
Assisted Living About
Service launched by TEPCO Retail – Aug Us
7th 2017
TEPCO Smart Home
“Keep your family secure even though you maybe far away”
Point 1: Notification when something “unusual” occurs
Point 2: Visualising the daily life of your loved ones on your smart phone
Point 3: “On demand” personal visiting service when you cannot (up to
visits per year included within your Plan)
Service fee: £20/month (24 month contract)
Potential Use Cases: 1. Home delivery service
In-home Prediction Technology Information for home delivery service
Probability of “somebody at home”
Lifystyle analysis for ads
16
Use Case 2. Activeness API for Nursing Care providers
Abnormal Life Pattern Detection Technology Assisted living app & robot
Estimates power usage pattern
and detects changes in usage pattern
Home care service
17
Use Case 3: "Model identification” API for marketing or Recall
Model Identification Technology Recall defective products
Identifies not only appliance type but also model number
Target marketing (appliance replacement)
18
Home IoT Platform Project
with
TEPCO Power Grid Company
Thank you
Taro.Tadano
President & CEO
Informetis Co.,Ltd
[email protected]
More Detail Introduction
about
Informetis Data Application
Range of Applications and Services
Various Applications
Collaboration with Partners
Able to efficiently use Able to realize what you Able to become more Able to confirm safety
energy didn’t notice in daily life convenient with partners when out of home
• Comparison • Activity detection • Real-time estimation • Abnormal detection
• Usage/Generation forecast • Lifestyle summary • Activity and usage detection • Appliance detection
• Energy saving tips • Abnormal detection • Activity and usage forecast • Lifestyle-based credit
• Tariff advice • For ads and sales • Remote control and • Activity forecast by human
• Device control integration • For logistics automation detection sensor
To Bigdata beyond energy data
23
Potential B2C Marketing applications
Provision of customer segmentation analysis based on appliance usage patterns and lifestyle data
Estimate of Households with
Extracting Lifestyle Patterns Estimating Household profiles
specific appliance type
An example of appliances detected
Air Rice Microwave
Conditioner Cooker
Washing Vacuum Kettle
Cleaner Hair Dryer TV
Machine
Promotion of Ready meals to Segmentation by daily appliance Eg. Single households, dual-
frequent Microwave users usage pattern income households, head of
household-head age group,
Estimation of vacation periods
Marketing campaign to homes single-person sex classification,
owning appliances which might be Detailed analysis of household etc.
due for change or recall activities(eg. Wake-up, Meal times, Estimate key changes in
TV viewing etc) for targeted household profile such as increase
marketing purposes and decrease of family size
24
Tracking your Home Life Log
Our Smart Sensor can track life events which could identify revenue opportunities at low cost
Cost of a traditional HEMS system is too high for consumers
Lower cost (around $2000 plus installation cost)
Compared to
traditional HEMS Hardware cost of our solution is significantly lower and so is the
on-going running cost of solution
Install a Single Smart Sensor
For Home/Building
New Build Easy to add to both new and existing homes (around 30
Vs minutes installation time)
Retrofit Can be offered as an “added value” service for rental properties
Palm-sized (60g) Unobtrusive Non-intrusive and unobtrusive monitoring of your home
Home Surveillance Lower rejection barrier by offering our “Home View” application
Other Further new and innovative applications currently being
developed by our engineers eg. Simple Home Security and
Applications
Assisted Living
25
Our Business Model
Basic model is one-off H/W cost plus on-going cost for access to disaggregated data via API
Basic Business Model
Various options available
Provided by Informetis
Provided by Partners
Waveform Data
Cloud 1
Integration of our API to your Platform
2
Service Provider
Residential Hardware cost, API for disaggregated
Home
Disaggregation Appliance Usage
Data data plus 2nd level support
Server
HEMS Service
3
Electricity Usage Forecast Office Value added services
Tariff Advice
Demand Response
4
Ads Total Power consumption plus
Store disaggregated data via API
Marketing, R&D
License
Light Security
5
Assisted Living
Sales of Smart Sensors
Smart Sensor Supplier Marketing Data Others
Smart Meter Supplier Smart Meter
or
Smart Sensor
*Future Plan
* Only residential homes service available at this time. Offices, stores and others available in the future
26
Back-up Slides
Typical Installation Images
This is an Engineer install and typically takes around 20-30 minutes
Informetis Smart Sensor Installation Example
The smart sensor can be placed into the
distribution board.
4 channel available
Mains (2ch) + Generation (2ch)
CT Inside the distribution board
Power cable connector
Average power consumption: 1W
Smart Sensor
28
