Scientists test AI that prevents wildfires by monitoring power lines

As wildfires continue to ravage the United States every year, destroying homes, threatening wildlife, and crippling infrastructure, many of them begin with damaged or downed power lines, so scientists are using artificial intelligence (AI) to help.

Specifically, these scientists believe that AI could detect sparks from power lines before they ignite flames, potentially saving lives and billions in damages, and have developed a suitable tool, per a report published by the National Renewable Energy Laboratory (NREL) on August 27.

As it happens, the culprit in many cases of a power-line related fire is what experts refer to as a high-impedance (HiZ) fault, which happens when a live wire touches the ground and creates a current, too weak for most systems to detect but strong enough to trigger sparks that can ignite a wildfire.

Stopping wildfires before they happen

To tackle this challenge, the U.S. Army Construction Engineering Research Laboratory (CERL) has partnered with the NREL to develop an AI-driven detection system. By training artificial neural networks – computational models designed to mimic the human brain – researchers have created a tool that can spot HiZ faults in real-time.

According to Richard Bryce, a senior researcher in power systems at NREL and lead on this project:

“The intention here is to enhance resilience in the power system and to enable faster responses during extreme events. (…) We want to provide utility companies with the tools for a more resilient power system with better reliability and security for customers that mitigates the potential for wildfires caused by high-impedance faults.”

Meanwhile, crucial to the effort was NERL’s collaboration with power management company Eaton, which simulated dozens of downed-conductor scenarios, accounting for ground conditions like grass or gravel, tree species, and moisture levels, to generate a wide dataset.

Then, the scientists fed this information into NREL’s advanced grid simulation platform, Power Systems Computer Aided Design (PSCAD), equipping machine learning models to identify even the tiniest fault signs. The result is an ensemble of AI models that utilities could deploy to monitor grids nationwide.

As the platform detects a potential fault, it allows for quick dispatching of crews, minimizing both outages and fire risks.