Top Ten Things Learned From Ten Years of Online Statistics Teaching (Michelle Everson)
IBCon Internet of Things: Ten Years of Lessons Learned
-
Upload
rob-hafernik -
Category
Data & Analytics
-
view
201 -
download
0
Transcript of IBCon Internet of Things: Ten Years of Lessons Learned
© Siemens Industry, Inc. 2015 All rights reserved. http://siemens.com/rcs
The Internet of Things - Ten Years of Lessons Learned
© Siemens Industry, Inc. 2015 All rights reserved. Building Technologies Division
The Site Controls platform helps retail chains achieve energy efficiency and smart consumption goals
Seven thousand installed sites with premium clients: ✓ Retail (small, medium, & big box) ✓ Grocery ✓ Restaurants ✓ Health Clubs ✓ Branch Banks
© Siemens Industry, Inc. 2015 All rights reserved. Building Technologies Division
Enterprise vs. "Building-Centric" approachSite Controls Enterprise Portal
On-demand exception management vs. alarm bombardment is one key
• Voluminous data isn't actionable; Information when prioritized is
• Identify most expensive enterprise problems first for best returns
• Verify corrective actions and service work remotely for all buildings
• Integrate & prioritize all energy consumption data into one platform
• "It's a Process not a Project”
Typical Customer: several hundred sites, several thousand HVAC units, several thousand lighting circuits, ten thousand sensors, a thousand energy meters
© Siemens Industry, Inc. 2015 All rights reserved. Building Technologies Division
Big Data!
• A typical site generates about 40,000 timestamped data points per day.
• That’s 40 million data points per day for an enterprise of one thousand sites.
• In a year… it’s too many to think about
• Without good analytics, the data is overwhelming
© Siemens Industry, Inc. 2015 All rights reserved. Building Technologies Division
Alerts and Alarms are not the answer
If only 1 in 10,000 data points causes an alert or alarm…
…and an enterprise generates 40 million data points per day:
4,000 Alerts or Alarms per day
“Alarm Fatigue”
http://en.wikipedia.org/wiki/Alarm_fatigue
© Siemens Industry, Inc. 2015 All rights reserved. Building Technologies Division
All of these buildings are in the Real World™
It’s a messy, complicated, uncooperative, unrelenting place.
Bad Conditions: things get smashed, painted over, disconnected and relocated
Bad Comms:
sometimes no comms at all, sometimes garbled comms
Bad Devices: sensors fail, sensors become uncalibrated, devices age
Human “Help”: humans are smart in an evil way
© Siemens Industry, Inc. 2015 All rights reserved. Building Technologies Division
All of these buildings are in the Real World™
It’s a messy, complicated, uncooperative, unrelenting place.
The natural environment is
harsh…
© Siemens Industry, Inc. 2015 All rights reserved. Building Technologies Division
All of these buildings are in the Real World™
It’s a messy, complicated, uncooperative, unrelenting place.
…not just harsh, but dangerous!
© Siemens Industry, Inc. 2015 All rights reserved. Building Technologies Division
All of these buildings are in the Real World™
It’s a messy, complicated, uncooperative, unrelenting place.
Things wear out in the Real World™, often
when no one is looking.
Fan Belt
© Siemens Industry, Inc. 2015 All rights reserved. Building Technologies Division
All of these buildings are in the Real World™
It’s a messy, complicated, uncooperative, unrelenting place.
Humans are the worst!
© Siemens Industry, Inc. 2015 All rights reserved. Building Technologies Division
All of these buildings are in the Real World™
It’s a messy, complicated, uncooperative, unrelenting place.
Lamp placed so as to heat thermostat
Humans are the worst!
© Siemens Industry, Inc. 2015 All rights reserved. Building Technologies Division
Plan for failure at every level
Controller/Gateway
Sensors/Devices
Wireless
Wired (Modbus)
Internet
Store
PrivateCorporateNetwork
VPN
Siemens Site
Controls
X
XX
X XX
XX
X - Point of failure
© Siemens Industry, Inc. 2015 All rights reserved. Building Technologies Division
Lucky us! The Real World™ runs on physics!
• Real-world physics limits the possible correlations in data
Example: a rooftop HVAC unit’s performance will not have any effect on the performance of a boiler in the basement
• This means that rules are more effective than statistics for analytics
Example: Outside Light Sensor should see regular periods of brightness and darkness. If not, sensor is likely broken, damaged, painted over or otherwise out of order.
Image by By Brews O’Hare (Own work)via Wikimedia
In Real World™ analytics, rules are usually handier than statistics
© Siemens Industry, Inc. 2015 All rights reserved. Building Technologies Division
Use data aggregation and ranking in your analytics toolkit
• Aggregation:
kWh / square foot / day Min kW, Max kW Average outside air temperature KPIs, eg, site connectivity
• Rules detect exceptions exceptions have scores
• Rank sites by aggregating exception scores in various categories:
HVAC Health Energy Use Site Connectivity On-site Comms Etc…
© Siemens Industry, Inc. 2015 All rights reserved. Building Technologies Division
Aggregate exception scores to rank sites
© Siemens Industry, Inc. 2015 All rights reserved. Building Technologies Division
All of these techniques used together lead us straight to the problems
HVAC not just broken, but blowing hot air
© Siemens Industry, Inc. 2015 All rights reserved. Building Technologies Division
Nature is tough
Human beings are worse more devious
Alerts and Alarms won’t cut it with Big Data
Plan for failure at every point
Rules are better than statistics in the Real World™
Aggregation reduces the amount of data you need to digest
Ranking makes the best or worst easy to find
Lessons71
2
3
4
5
6
7
© Siemens Industry, Inc. 2015 All rights reserved. Building Technologies Division
http://siemens.com/[email protected]@robhafernik