EDF Energy R&D UK Centre
Transforming the energy system with AI
David Ferguson,
Head of Digital Innovation
The rules of digital innovation
Move fast and break things.
Unless you are breaking stuff,
you are not moving fast
enough.
“
”- Mark Zuckerberg, Facebook
Move fast and break things?
The old energy system
The new energy system
The new energy system… it’s really complex!
1 million homes with
solar panels
1 million homes with
smart thermostats
Energy innovation is not coming from the old energy sector
New tech + new ways of working
80
5 Teams People
40 Phd students15 NationalitiesBudget
Nuclear
Renewables
Smart Customers
Digital Innovation
Open Innovation
Brighton
Manchester
London
£40m
Conventions or contracts with EDF
Energy Business Units or EDF Group100% 26% 10%
Public fundingCorporate activities
Funding
3 Locations
EDF Energy R&D UK Centre
Image: Wired
- Andrew Ng, Baidu, 2016
AI will transform many
industries.
But it’s not magic.
“
”
Let’s not get carried away just yet
What is it? Use cases for EDF Energy
Generation
● Condition monitoring
● Predictive maintenance
● Digital replicas
● Operational support
● Autonomous robotic devices
Customers
● Customer service
● Energy use analytics
● Home automation
● Optimisation and trading
Grids and networks
● Predictive maintenance
● Demand-side response
● Trading
Support functions
● Supply chain optimisation
● Legal services
Reasoning
Knowledge
Planning
Learning
Natural Language
Understanding
Perception
How could AI apply to EDF Energy?
Early R&D
R&D UK Centre
Digital replicas: label recognition in power stations
Images: EDF
30,000
labels
500 bn
pixels
4
person-months
per building
10
buildings per
year
Digits zone cropping
Preprocessing (B&W and cleaning)
Digits recognition:
02615.0
Customer service: optical character recognition
Energy use analytics: Virtual Data Lab
Customer service: chatbots and voice
Our vision
R&D UK Centre
Real-time condition
monitoring
Two use cases for AI for EDF Energy
The intelligent home
Real-time condition
monitoring
AI in power stations
1 - Explain
2 - Predict
3 - Prescribe
4 - Automate
Real-time condition
monitoring
AI for residential customers
The intelligent home
1 - Show me
2 - Help me
3 - Do it for me
The challenges of turning vision into reality
R&D UK Centre
Delivering the promise of AI in a ‘legacy’ company
Challenges
Generic
Specific to legacy
orgs.
Skills
Ethics
Privacy
Security
Data
Connectivity
Roll out
‘The day job’
Conclusions
 The energy system is complex and is changing
 AI is not magic
 It will help manage the new, complex system
 Our approach: start with small, difficult problems
 Legacy companies have a significant disadvantage
Thank you!
david.ferguson@edfenergy.com
@daveyf
R&D UK Centre

AI in Energy

  • 1.
    EDF Energy R&DUK Centre Transforming the energy system with AI David Ferguson, Head of Digital Innovation
  • 2.
    The rules ofdigital innovation Move fast and break things. Unless you are breaking stuff, you are not moving fast enough. “ ”- Mark Zuckerberg, Facebook
  • 3.
    Move fast andbreak things?
  • 4.
  • 5.
  • 6.
    The new energysystem… it’s really complex! 1 million homes with solar panels 1 million homes with smart thermostats
  • 7.
    Energy innovation isnot coming from the old energy sector New tech + new ways of working
  • 8.
    80 5 Teams People 40Phd students15 NationalitiesBudget Nuclear Renewables Smart Customers Digital Innovation Open Innovation Brighton Manchester London £40m Conventions or contracts with EDF Energy Business Units or EDF Group100% 26% 10% Public fundingCorporate activities Funding 3 Locations EDF Energy R&D UK Centre
  • 9.
    Image: Wired - AndrewNg, Baidu, 2016 AI will transform many industries. But it’s not magic. “ ” Let’s not get carried away just yet
  • 10.
    What is it?Use cases for EDF Energy Generation ● Condition monitoring ● Predictive maintenance ● Digital replicas ● Operational support ● Autonomous robotic devices Customers ● Customer service ● Energy use analytics ● Home automation ● Optimisation and trading Grids and networks ● Predictive maintenance ● Demand-side response ● Trading Support functions ● Supply chain optimisation ● Legal services Reasoning Knowledge Planning Learning Natural Language Understanding Perception How could AI apply to EDF Energy?
  • 11.
  • 12.
    Digital replicas: labelrecognition in power stations Images: EDF 30,000 labels 500 bn pixels 4 person-months per building 10 buildings per year
  • 13.
    Digits zone cropping Preprocessing(B&W and cleaning) Digits recognition: 02615.0 Customer service: optical character recognition
  • 14.
    Energy use analytics:Virtual Data Lab
  • 15.
  • 16.
  • 17.
    Real-time condition monitoring Two usecases for AI for EDF Energy The intelligent home
  • 18.
    Real-time condition monitoring AI inpower stations 1 - Explain 2 - Predict 3 - Prescribe 4 - Automate
  • 19.
    Real-time condition monitoring AI forresidential customers The intelligent home 1 - Show me 2 - Help me 3 - Do it for me
  • 20.
    The challenges ofturning vision into reality R&D UK Centre
  • 21.
    Delivering the promiseof AI in a ‘legacy’ company Challenges Generic Specific to legacy orgs. Skills Ethics Privacy Security Data Connectivity Roll out ‘The day job’
  • 22.
    Conclusions  The energysystem is complex and is changing  AI is not magic  It will help manage the new, complex system  Our approach: start with small, difficult problems  Legacy companies have a significant disadvantage
  • 23.