Pacific Northwest National Laboratory has developed an artificial intelligence tool that aims to help utilities analyze the reams of data used to track the status of the electric power grid.
PNNL researchers created the program, ChatGrid, so a grid operator can ask a question about the grid and get an easy-to-interpret answer.
The researchers were responding to concerns raised by utility colleagues who have to comb through data about regional grids and refer to visualizations of which power plants are generating how much energy and where that energy is flowing in order to make quick decisions when there is a disruption caused by a storm or equipment failure.
Those “tools can be cumbersome and navigating them can slow down decision-making,” Shrirang Abhyankar, an optimization and grid modeling researcher at PNNL, said in a statement.
ChatGrid runs on a publicly available large language model, which works somewhat like the predictive text on a smartphone. In the training process, the program is fed massive amounts of text -- English, in this case – so it can recognize questions or commands and supply answers it has deemed statistically relevant.
But, because it is highly sensitive, the PNNL researchers did not train the program on grid infrastructure data. Instead, they compiled grid infrastructure data into their own internal database with columns for data such as “capacity” or “location” of power plants.
Then, rather than training the program on the data itself, it is trained to know there are columns with labels so that “ChatGrid can still produce grid visualizations while keeping the nation’s grid data safe,” the researchers said.
To further protect the safety of grid data, ChatGrid’s visualizations do not currently represent real-life grid data, the researchers said. It uses synthesized data from the Exascale Grid Optimization model developed by PNNL, four other national labs and Stanford University. The ExaGO model can simulate the nation’s power grid in real time, allowing grid planners to analyze the ripple effects of any disruptions.
ChatGrid is available for download on GitHub. “We'd really like to put this technology in front of the operators and let them input questions and get feedback to see how ChatGrid is performing,” Abhyankar said.
After receiving feedback from users, Abhyankar said he hopes to build a better version of ChatGrid that grid operators can use safely in their control rooms with real-life data. For that to work, however, ExaGO’s massive datasets need to be accessible to regular computers.
“With ChatGrid, we can translate this data into something that's actionable to a human,” Chris Oehmen a computational biologist at PNNL, said in a statement. “It’s a first really important step in letting grid operators interface with those big datasets in a way that's intuitive.”
“We’re envisioning a new way to look at data through questions,” Abhyankar said. “ChatGrid allows someone to query the data—in a literal sense—and get an instantaneous answer.”
