AI-enhanced early detection of electrical grid arcing to help prevent wildfires
Published on: June 22, 2026
Researchers at the U.S. Department of Energy’s Oak Ridge National Laboratory (ORNL) have unveiled an AI-powered platform designed to detect abnormal electrical grid conditions—especially subtle arcing faults—that are often invisible to traditional sensors. These arcing events can spark wildfires, damage electrical infrastructure, and trigger power outages.
The new system rapidly analyzes waveform data from grid monitoring equipment using advanced signal‑processing techniques enhanced by AI. It magnifies faint signatures of arcing faults—raising waveform visibility from approximately 6 percent to 72 percent in tests—making previously hidden disturbances detectable.
ORNL is partnering with Southern California Edison (SCE) to test the system using five years of real-world grid data. The goal is to refine detection accuracy, speed, and sensitivity under real operating conditions and integrate the technology into SCE’s internal analytics infrastructure for practical deployment.
Beyond detecting arcing, the platform classifies seven types of electrical fault conditions—including overcurrent faults, blown fuses, motor starts, and capacitor switching—enhancing utilities’ situational awareness and enabling preventive maintenance.
ORNL plans to make the detection tools available in an open‑source “data analytics toolbox” linked to its Grid Event Signature Library, allowing utilities and researchers nationwide to leverage the algorithms freely in efforts to enhance grid safety and resilience.
This development comes amid growing concerns over wildfire risks driven by electrical faults and severe weather. By enabling early identification of dangerous grid anomalies, the AI system promises to improve utility response times and help mitigate the significant human and economic costs associated with wildfire-related outages.
No comments yet.