IEEE Electrification - December 2022 - 57

possible. There are APIs available for utilities data, solar
panel systems, and building automation systems.
The Internet of Things
The concept of the Internet of Things (IoT) refers to the
increased connectivity of devices. The increased popularity
of streaming data from enabled devices has lessons
that can be applied to BTM DERs as well. Highly connected
devices enable remote asset monitoring, predictive maintenance,
and device management. Many analytical software
tools have built-in functionality for handling IoT
data and performing streaming analytics. Examples
include SAS Event Stream Processing, Amazon Kinesis,
Microsoft Azure IoT Hub, and Google Cloud IoT.
Use Cases and Data Handling for Streaming Data
With the increasing volume of data from operations,
transactions, sensors, and IoT devices, it has become necessary
to analyze data in real time to obtain useful
insights, make sound decisions, and scale from edge to
cloud as the data volume grows. There is now a wide
breadth of analytical tools and methods for implementing
high-frequency analytics. These tools connect, decipher,
cleanse, and generally process streaming data. The output
of these tools can be standardized labeled data that can be
stored in a data warehouse or less curated data that lands
in a data lake.
Distributed, shared-memory grid hardware and
cloud environments provide faster and more powerful
stream processing. Advanced analytics with embedded
AI and machine learning capabilities analyze structured
and unstructured data sources-including text, images,
and video-as they stream. Algorithms like support vector
data description, robust principal component analysis,
random forest, gradient boosting, and streaming
regression analysis are being applied in stream.
An example of streaming data analytics is at commercial
solar farms. Utilities that incorporate large-scale DERs,
such as solar farms, must plan for the intermittency of
these resources. Streaming data from solar farms that
include current solar production and current weather conditions
(the cloud cover and irradiance) can help inform
forecasting models that will determine how much solar
generation to expect in the next hour or even the next few
minutes. Utilities and grid operators can then plan how
much energy to purchase to meet demand.
Conclusion
Our desire for awareness, control, and connectivity combined
with the need for a greener planet are forcing the
transition of the electric grid to a democratized ecosystem
with dynamic exchange of energy and information. Energy
resources are increasingly distributed. Synchrophasors,
AMI meters, sensors, and control equipment and devices
for measuring and managing the exchange of energy produce
a high velocity of data. These big data need to be
governed, stored, and analyzed as they occur, oftentimes
at the edge of the grid and beyond the point of the utilities'
visibility and supervision.
The prevalence of installed sensors, access to high-frequency
data, and analytic software advancements have
created opportunities for the real-time monitoring and
analysis of DERs. Advancements in analytics and computing
hardware have grown in parallel with these developments
in the utilities industry. Analytic models and
machine learning models help estimate the amount of
BTM energy production, capturing the relationship
between weather conditions and known DER production
while incorporating weather forecasts to estimate the
expected production. Analytic software facilitates the fast
development and deployment of site-specific models (such
as one customized model per solar farm or wind farm).
The advancement of computing, accessibility of computing
and storage, and access to big data coupled with
newfound analytical techniques are turning DER data into
knowledge. The knowledge gained from their study can be
reapplied in other settings to assist energy providers in
DER and net load forecasting to fill in gaps where some
but not all variables and data are known. The proliferation
of DERs calls for the understanding, prediction, and study
of how they relate to the grid and production. The challenges
that come alongside the benefits of BTM DERs can
be addressed with the tools and techniques outlined in
this article. New smart data sources coupled with powerful
hardware and advanced analytics, are an opportunity
to create resilient and well-planned grids that intelligently
incorporate DERs.
For Further Reading
" How to do deep learning with SAS, " SAS Inst. Inc., Cary, NC,
USA, 2019. [Online]. Available: https://www.sas.com/content/
dam/SAS/en_us/doc/whitepaper1/deep-learning-with
-sas-109610.pdf
R. Kaur and S. Chopra, " Virtualization in cloud computing:
A review, " Int. J. Sci. Res. Comput. Sci., Eng. Inf. Technol., vol. 6,
no. 4, pp. 1-5, Jul./Aug. 2020, doi: 10.32628/CSEIT20641.
U. A. Khan, W. U. Bajwa, A. Nedic´, M. G. Rabbat, and A. H.
Sayed, " Optimization for data-driven learning and control, "
Proc. IEEE, vol. 108, no. 11, pp. 1863-1868, Nov. 2020, doi:
10.1109/JPROC.2020.3031225.
Biographies
Arnie de Castro (arnie.decastro@sas.com) is with SAS
Institute, Cary, NC 27513 USA, a software company that
specializes in analytics, artificial intelligence and data
management.
Ashley Mui (ashley.mui@arcadia.com) is with Arcadia,
DC 20004 USA, a climate technology company that built
an application programming interface for global energy
data and manages 1 GW of community solar.
Garrett Frere (garrett.frere@snowflake.com) is with
Snowflake Bozeman, MT 59715, USA, a cloud computing
data platform.
IEEE Electrification Magazine / DECEMBER 2022
57
https://www.sas.com/content/dam/SAS/en_us/doc/whitepaper1/deep-learning-with-sas-109610.pdf https://www.sas.com/content/dam/SAS/en_us/doc/whitepaper1/deep-learning-with-sas-109610.pdf https://www.sas.com/content/dam/SAS/en_us/doc/whitepaper1/deep-learning-with-sas-109610.pdf
IEEE Electrification - December 2022

Table of Contents for the Digital Edition of IEEE Electrification - December 2022
