As research tools like computers or microscopes become more and more powerful and sophisticated, the amount of information they can gather is tremendous. Today, we have machines and tools that can collect, analyze, and decode enormous amounts of data.

Modern scientific facilities, instruments, and high-performance computing tools can interpret a volume of data that traditional analytical methods struggle with—for example, electron microscopes can now generate a terabyte, or one million bytes, of data in a single experiment.

These tools have been used to determine how to design solar cells that can absorb light more efficiently or how captured carbon dioxide can be converted into usable products—and these are just a few examples. This explosion of accessible data has helped scientists and researchers unlock the full potential behind all this information.

DOE is using artificial intelligence (AI) and data monitoring tools to inform and shape the future of our critical infrastructure. Faster grid analytics and modeling, better grid asset management, and sub-second automatic control actions will help system operators avoid power outages, improve operations, and reduce costs.

In April 2019, DOE announced a $7M investment to explore the use of big data, AI, and machine learning technology and tools to derive more value from the vast amounts of sensor data already being collected to monitor the health of the electric grid and support system operations.

Projects selected for this investment were provided with a pre-determined dataset to use in their analysis. On July 28, 2021, four of these project teams will share their findings during a public webinar.

AI and machine learning techniques enable users to “see” grid behavior that is nearly imperceptible to a human operator or system planner. Learning about these behaviors contributes to improved reliability, resilience, and system understanding in a rapidly changing context. Despite over 100 years of engineered electric power systems, these big data projects have highlighted how far technology has come, while showing how far technology needs to go to truly understand the science behind grid behavior. The project teams’ work features:

  • Applying signature identification to a large amount of electric grid data—streaming or historical—to quantify the relative severity, location, and duration of grid events.
  • Applying new techniques to every stage of the machine learning pipeline, from preprocessing to event detection, to provide increased situational awareness in an online environment.
  • Using innovative labeling methods for data to uncover the most value and apply semi-supervised learning approaches to relevant grid events.
  • Integrating state-of-the-art sensor analytics and machine learning into real-time grid monitoring to develop a robust event diagnostics platform.

All these projects, and projects like these, have the potential to reveal critical new insights in research that can help tackle clean energy, climate, and national security challenges for the American people. AI is already shaping our lives in countless ways, and its impacts will only become more significant.