A National Infrastructure for Artificial Intelligence on the Grid
This paper overviews a three-year Open Innovation project funded by the Advanced Research Project Agency - Energy (ARPA-E) to launch a national infrastructure for artificial intelligence (NI4AI) for the electric grid. The national infrastructure will be composed of three major components: high-value datasets from grid sensors for research and development, an open, accessible, cloud-based AI Platform designed for time series sensor data, and a community of people interested in advancing analytics for the grid. The project aims to accelerate the development of analytics, machine learning, and AI to improve all aspects of the power grid.
The first component of the project will deploy a variety of high-frequency grid sensors across the power system. The sensors, including synchrophasors, point-on-wave sensors, and more, will be used to capture both wide-area data and local system data. These sensors, in turn will generate high-value datasets suitable for research and development purposes.
The second aspect of this project is the deployment of a cloud-based Artificial Intelligence platform. The platform will be able to consume the time series data provided by the sensor network. In addition, other research data sets will also be imported and made available. More importantly, the platform will serve as a foundation wherein analytics can be developed, and AI techniques can be leveraged on the data.
Finally, the project aims to create an interdisciplinary community of practitioners and subject matter experts both within and outside of power engineering. This community will be developed through shared content and code, conferences and workshops, and data science competitions, hopefully resulting in in an open and collaborative ecosystem.