GMLC 1.4.9 – Integrated Multi Scale Machine Learning

You are here

Lead Performer: Lawrence Berkeley National Laboratory – Berkeley, CA
-- Los Alamos National Laboratory – Los Alamos, NM
-- Lawrence Livermore National Laboratory – Livermore, CA
-- National Renewable Energy Laboratory – Golden, CO
-- Oak Ridge National Laboratory – Oak Ridge, TN
-- Sandia National Laboratory – Albuquerque, NM
-- Power Standards Lab – Alameda, California
-- National Instruments – Austin, TX
-- PingThings – San Juan, CA
-- Riverside Public Utility – Riverside, CA
-- Florida Power and Light – Juno Beach, FL
DOE Total Funding: $3,730,000 (funding is a joint EERE/OE initiative, BTO funds contribute to total)
Cost Share:
Project Term: 1/2016-3/2019
Funding Type: DOE Grid Modernization Laboratory Consortium (GMLC) Lab Call

Project Objective

As a part of the Department of Energy’s Grid Modernization Initiative, the Grid Modernization Laboratory Consortium projects represents a comprehensive portfolio of critical research and development in advanced storage systems, clean energy integration, standards and test procedures, and a number of other key grid modernization areas.

This project’s overarching goal is to create advanced, distributed data analytics capability within the DOE GM Consortium, to provide visibility, and controllability to distribution grid and building operators.   While there has been significant development of analytics methods with streaming data at the transmission level, distribution and buildings level analytics are in the elementary stages. Machine learning is the basis for the analysis development, and will allow new levels of visibility and resource integration to be achieved using open and utility datasets at both the building and distribution level.  This project will develop the necessary framework to promote and integrate a larger distributed analytics activity utilizing existing and new data sources developed through both private and DOE partnerships.

The project will seek to demonstrate the full capability of synchronized disparate data sources for distribution and building grid analysis and control with application of machine learning, enabling future distributed applications such as transactional energy validation, fault analytics and failure prediction, resilience applications and a BMS/DMS integration methodology for transition to industry. These use cases are being identified as having multi program application, along with a key position in the modernized grid where advanced analytics could be transformational to performance and accuracy.

Project Impact

This project will evaluate and validate the application of machine learning techniques to create actionable information for grid and building operators, and derive customer benefits from disparate data.


DOE Technology Manager: Joe Hagerman
Lead Performer: Emma Stewart, Lawrence Berkeley National Laboratory