Upload
hoangtram
View
226
Download
3
Embed Size (px)
Citation preview
October 2016 Industrial Internet of Things
© Dimitry Gorinevsky 1
Analytics for Industrial Internet of Things
Dimitry Gorinevsky Consulting Professor in Electrical Engineering
Stanford University www.stanford.edu/~gorin
CEO, Mitek Analytics LLC [email protected]
NAS GUIRR The Fourth Industrial Revolution ● October 25, 2016
www.mitekan.com
INDUSTRIAL REVOLUTION IIoT Analytics
October 2016 Industrial Internet of Things
© Dimitry Gorinevsky 2
Digital Revolution
• Software is eating the world • (Marc Andreessen, 2011)
• Internet Revolution
– Connected people
October 2016 Industrial Internet of Things
© Dimitry Gorinevsky 3
New Industrial Revolution
• Digital revolution
– connected people
– 10-15% of the economy
• Industrial IoT revolution
– Connected machines
– 80% of the economy
October 2016 Industrial Internet of Things
© Dimitry Gorinevsky 4
More Than Manufacturing
• Development and manufacturing
– 10-15% of the lifecycle cost
• The IIoT will change the operations and support
– 65-80% of the lifecycle cost
October 2016 Industrial Internet of Things
© Dimitry Gorinevsky 5
DoD CBM+ 2007
Value chain for industrial goods (PWC, 2015) Manufacturing
Business Value Estimates
• Analyses of the IIoT economic impact
October 2016 Industrial Internet of Things
© Dimitry Gorinevsky 6
Value Date Comment
GE $10-15 Trillion 2014 IIoT
Accenture $14 Trillion 2015 IIoT
McKinsey $11 Trillion 2015 IIoT
Industrie 4.0 $4 Trillion 2014 Manufacturing
Gartner $2 Trillion 2015 Consumer IoT
Cisco $17 Trillion 2015 IoE IIoT+CIoT
IIOT ANALYTICS IIoT Analytics
October 2016 Industrial Internet of Things
© Dimitry Gorinevsky 7
Enterprise Architecture View
• IIoT Applications – Analytics: process and
analyze the data
– Operations: business processes
• IIoT Platform – Collect and manage data
– Needed to run applications
– Most action, so far
October 2016 Industrial Internet of Things
© Dimitry Gorinevsky 8
Data Architecture
Technology Architecture
Business Architecture
Applications Architecture
Technology Investment
Added Value
IIoT Data Analytics
• IIoT = IT systems using big OT data
• In IIoT, the OT data is accumulated as Persistent Data
• OT systems use their data on-line, but do not accumulate it
October 2016 Industrial Internet of Things
© Dimitry Gorinevsky 9
Operational Technology
Persistent Data
Operations Research
Control
Signal Processing
IIoT Analytics Data Science
Information Technology
EXAMPLES OF IIOT ANALYTICS APPLICATIONS
IIoT Analytics
October 2016 Industrial Internet of Things
© Dimitry Gorinevsky 10
IIoT Analytics Stack
October 2016 Industrial Internet of Things
© Dimitry Gorinevsky 11
Embedded Control and Optimization
IT
OT
Digital Twin
Analytics
Planning and Strategy
Risk Analytics
Demand Prediction
Prescriptive Analytics
Operational Digital Twin
Air Force, NASA Mitek, Stanford GE, Airlines, …
October 2016 Industrial Internet of Things
© Dimitry Gorinevsky 12
Predictive Analytics
Train Aircraft and Engine Models
Flight Data Repository Operational Data
Fleet Performance Models (Digital Twin)
Fuel Efficiency
BI Reports
Aircraft Fleet
FDRs
10 Tb 2 Gb 12 hours
Fault Root Causes Condition Trends
IT
OT
LIMS-EV
Demand Prediction
Prescriptive Analytics
Predictive Analytics
Data Repository
Fleet Reliability Models (Digital Twin)
Report bad actors Report bad hosts
Demand Prediction for Logistics
Failure Event Data
Train Reliability Model
Maintenance
OO-ALC, Hill AFB Mitek Lockheed Martin
GCSS-AF
October 2016 Industrial Internet of Things
© Dimitry Gorinevsky 13
IT
OT
Prescriptive Analytics
Weather Risk Analytics
October 2016 Industrial Internet of Things
© Dimitry Gorinevsky 14
Predictive Analytics
Extreme Weather Risk Model Training
Weather Data Repository
Baseline; extreme anomalies Risk Assessment Reports: • Insurance • Utilities
Data Feeds
• Year-to-year trends of 100-year event risks – Joint work with Steven Chu, Stanford Physics, and NCAR
• High temperature risk increased x2 in last 40 years
Bulk Power System Planning
• Planning for reliable power
– Load loss < 1 day in 10 years
– Data driven probabilistic model for load balancing
October 2016 Industrial Internet of Things
© Dimitry Gorinevsky 15
Loss Of Load risk
system load
CONCLUSIONS IIoT Analytics
October 2016 Industrial Internet of Things
© Dimitry Gorinevsky 16
Conclusions
• The IIoT will create value through data-driven analytics applications
– Convergence of IT and OT computing
• Value in operations - not just manufacturing
• Many layers of analytics stack are important
• Examples of the IIoT analytics
– Strategy: risk analytics
– Logistics: demand prediction
– Asset fleet analytics: Digital Twin
October 2016 Industrial Internet of Things
© Dimitry Gorinevsky 17