A National Infrastructure for Artificial Intelligence on the Grid

Executive Summary

This proposed national infrastructure for artificial intelligence on the grid is composed of three components:

  1. a high performance and open platform for sensor data storage, access, and analysis that enables collaboration between multiple organizations and is used both in research and in production within utilities
  2. a hyperscale set of open sensor data (both real and synthetic) that captures different aspects of the electric grid’s behavior, and
  3. a set of structural incentives providing motivation for both researchers and industry practitioners to collaborate.

The all-important data will be created by (a) deploying two groups of sensors to create both a wide area, continuous data set and boutique data sets targeting grid components of interest, (b) capturing data from virtual sensors instrumenting large scale grid models, and (c) creating a data escrow service that facilitates the exchange of sensor data between utilities and pre-approved organizations, complete with standardized data cleaning, optional obfuscation, and legal documentation. The overarching objective is to remove any and all obstacles to the rapid development, adoption, and deployment of new use cases based on analytics, machine learning (ML), and artificial intelligence (AI) for sensor data measuring the electric grid.