Researchers have developed an artificial intelligence model that automatically reroutes electricity in milliseconds to help power grids prevent blackouts.
The approach is an early example of “self-healing grid” technology, which uses AI to autonomously detect and repair problems such as outages without human intervention when problems occur, such as from storms. Damaged power lines.
Although more research is needed before the system can be implemented and scaled to a real-world power grid, it is an exciting development for the country's troubled power grid, the researchers say.
Co-corresponding author Soma Chowdhury, associate professor in the University at Buffalo's Department of Mechanical and Aerospace Engineering, says, “Power grids around the world are being challenged by an increasing number of extreme weather events, the potential for cyber-attacks, and an expected increase in demand. being challenged.” .
“Therefore, it is important that we develop tools that modernize the system and make it more resilient against future power outages.”
Chaudhary is the co-director of the Center for Embodied Autonomy and Robotics (CEAR).
The North American grid is a vast, complex network of transmission and distribution lines, generation facilities, and transformers that distribute power from power sources to consumers.
Using various scenarios in test networks, the research team demonstrated that its solution can automatically identify alternative routes to transfer electricity to customers before an outage occurs. Once trained, AI has a speed advantage: the system can automatically reroute electrical flow in microseconds, whereas current processes involve classical engineering techniques — or human intervention — determining alternative routes. This can take anywhere from minutes to hours.
“Our goal is to find the best way to send electricity to as many customers as quickly as possible,” says co-corresponding author Ji Zhang, an associate professor of mechanical engineering in the Eric Johnson School of Engineering and Computer Science at UT Dallas. Is a professor.
To map the complex relationships between the entities that make up the power distribution network, the research team used algorithms that apply machine learning to the graph. In this context, graph machine learning involves describing the topology of a network, the way different components are arranged in relation to, or in relation to, each other and how electricity moves through the system.
The team also relied on reinforcement learning — where a virtual agent is typically deployed in a simulated environment of a real problem — to systematically present scenarios and gradually learn from that experience. An example of knowledge gained from such experience would be if there is a power outage due to line faults. The system would then be able to reset using switches and draw power from nearby sources, such as large-scale solar panels or batteries on a university campus or business.
“These are decisions that the model can make almost instantaneously, which has the potential to eliminate or greatly reduce the severity of power outages,” says co-first author Steve Paul, who led the study. Worked on this project earlier in the year while pursuing a Ph.D. . Paul is now a postdoctoral scholar at the University of Connecticut.
After focusing on preventing outages, researchers now want to develop similar technology to repair and restore the grid after power outages, such as those caused by natural hazards.
It appears in the research. Nature Communications.
Additional authors are from the University of Texas at Dallas.
This work was supported by the US Office of Naval Research and the National Science Foundation.
Source: University at Buffalo