WVU researchers are utilizing a DOE grant to develop machine learning to fortify power grids

Student working in Smart Grid Resiliency and Analytics Lab (WVU Photo/Brian Persinger)

When the weather gets bad, researchers at West Virginia University have underserved communities in mind, especially during a power outage.

Story by Kaley LaQuea, Communications Specialist
Photos by Brian Persinger

Benjamin M. Statler College of Engineering and Mineral Resources

A team of researchers from the Statler College of Engineering and Mineral Resources will receive $150,000 in funding as part of a Department of Energy grant for the Renewable Advancing Community Energy Resilience (RACER) Program to enhance power grid resilience for underserved communities that are more vulnerable following power outages caused by extreme weather events.

Lane Department of Computer Science and Electrical Engineering chair Anurag Srivastava will lead the WVU team that’s developing algorithms using a machine-learning approach to estimate energy resources behind the meter and cold load pick up for power restoration.

“At WVU, we're excited to be part of the DOE RACER project, working in collaboration with a dedicated team led by Washington State University,” Srivastava said. “With our focus on using machine learning to estimate behind-the-meter Distributed Energy Resources, our goal is to enhance power grid resilience, especially in underserved communities. We hope to mitigate the adverse impacts of power outages on vulnerable communities, ultimately helping utilities provide more reliable power to those who need it most."

The project, led by a WSU team includes collaborations with Pacific Northwest National Laboratory and Illinois electric utility provider ComEd. Researchers will work with community members in Rockford, IL to forge solutions for power outages as they occur due to extreme weather events.

Researchers will enhance power grids through solar photovoltaic (PV) energy and storage systems at field testing sites. The team will conduct data analysis and modeling to install devices to existing structures to be able to remotely manage the grid and anticipate outages and power needs, especially in the event of a natural disaster and extreme weather.

Underserved communities are disproportionately impacted by power outages. Communities like Rockford, which comprises a high percentage of low-income households and individuals from marginalized backgrounds, have fewer resources to adapt when power outages occur.

Srivastava says that certain structures might be inaccessible after storms or natural disasters and grid operators aren’t able to cover large geographic areas as quickly or effectively.

“For someone to physically go and get key field information is challenging, so, if possible, it should be done at the central location or operation center or at another data aggregation point,” Srivastava said. “And what we’ll do here is applicable to any disadvantaged community where there is not enough redundancy in the circuits to feed the power supply which feeds that community.”

---