[Claude Godin]

Hi All

 

We did two interesting end use projects in the middle east where we had to use non-intrusive metering for cultural reasons  to develop end use load shapes for input new bottom up forecasting methods to compliment the current top-down approach .  The first was for electricity  where we used 1,200 commercially available measurement instruments that provided high frequency measurements and edge detection at the meter level. The load and edge data from the instruments were transmitted to a central station on a daily basis  where cluster analysis was performed to develop load signatures, which were them classified in specific appliances using the vendor’s load categorization algorithms.   

 

The second was for water, where no commercially available NILM equipment was available.  First,  a pilot study was done where 250 sites where instrumented with standard flow meters that were configured to produce pulse outputs at a rate of 0.05/liters per pulse (about ½ shot glass resolution) , which were fed into purpose-built data loggers that recorded data a sub-second rate.  Data was transmitted  daily via GPRS modems to a central station.  After three months of data collection, the data was semi-automated processed into  end use consumption patterns.   Although the semi-automated   process produced great results for the pilot, it was labor intensive, and the national roll-out required that 2,500 residences be instrumented and monitored for a one year period-clearly a better method was required over the semi-automated process!

 

The attached presentation discusses the drivers for going behind the meter, a brief description of the electric system, and then tells the story how we explored several machine learning techniques in combination with the pilot data as ground truth and additional survey data to process the 100,000,000,000 rows of flow data into end use flow profiles for our new water bottom-up forecast.

 

In the end the end,  several methods were tested for classification, but the one that stood out was Convolutional Neural Network pattern recognition.   

 

 

 

Sincerely,

Claude Godin

Senior Principal Consultant, Policy Advisory and Research, Key Initiatives, PAR

 

DNV GL Energy Insights USA, Inc.
claude.godin@dnvgl.com
Mobile +1 (919) 539-3231  

dnvgl.com  |  LinkedIn

DNV GL® – a member of Det Norske Veritas® group.

 

**************************************************************************************
This e-mail and any attachments thereto may contain confidential information and/or information protected by intellectual property rights for the exclusive attention of the intended addressees named above. If you have received this transmission in error, please immediately notify the sender by return e-mail and delete this message and its attachments. Unauthorized use, copying or further full or partial distribution of this e-mail or its contents is prohibited.
**************************************************************************************

——Original Message——

Hello everyone,

An interesting question was posted to one of our Analytics Communities and I was hoping to re share here. Please see below and let us know what input you may have!

“​​Is there member (non vendor) who has a list of vendors offering platforms for advanced pattern recognition?”

Thanks!

——————————
Kevin Praet
Membership Coordinator
Informa
Boulder CO
315-440-3033
——————————

[Liliana Wilson]

​Hello,

At Enbridge we are experimenting with Azure Cognitive Services – the ‘Vision” Services are pretty good at Pattern Recognition .

https://azure.microsoft.com/en-ca/services/cognitive-services/

In addition, the out-of-the-box Azure Cognitive Services can be enhanced with Custom skills, proving an opportunity to improve the accuracy.

Give it a try.

Regards,
Liliana Wilson

——————————
Liliana Wilson
Supervisor Big Data Analytics
Enbridge Gas Inc
North York ON
416-495-5858
——————————
——————————————-
Original Message:
Sent: 08-15-2019 16:38
From: Kevin Praet
Subject: Advanced Pattern Recognition

Hello everyone,

An interesting question was posted to one of our Analytics Communities and I was hoping to re share here. Please see below and let us know what input you may have!

“​​Is there member (non vendor) who has a list of vendors offering platforms for advanced pattern recognition?”

Thanks!

——————————
Kevin Praet
Membership Coordinator
Informa
Boulder CO
315-440-3033
——————————

[Kevin Praet (Adm)]

Hello everyone,

This is just a reminder for our webinar next week Thursday (8/29) “Taking the Risk out of Risk Management,” with member utility TPU. For more information and registration, please visit our webinar page here and let us know if you have any questions.

Thanks!

——————————
Kevin Praet
Membership Coordinator
Informa
Boulder CO
315-440-3033
——————————
——————————————-
Original Message:
Sent: 08-08-2019 16:50
From: Kevin Praet
Subject: Another August Webinar! (August 29th) – “Taking the Risk out of Risk Management”

Learn to “take the risk out of risk management” in this exciting webinar with TPU! For more info and registration, please visit our webinar page here.

Aging infrastructure is a top concern for utility companies today, and the ability to strategically prioritize work is more imperative than ever in order to minimize risk profiles. In this member-only webinar, Todd Shepherd, Strategic Asset Management Supervisor at Tacoma Power, will share how his group uses a Bayesian approach to quantifying the level of uncertainty in particular asset failure scenarios, and how that influences its overall corporate risk profile. He will then describe how that information is then used to drive data and asset inspection strategies needed to lower TPU’s overall risk exposure as a company. 

We hope to see you all there!

——————————
Kevin Praet
Membership Coordinator
Informa
Boulder CO
315-440-3033
——————————