A key component of PG&E's wildfire mitigation strategy involves collaboration with Schweitzer Engineering Laboratories (SEL) to develop high impedance fault (HIF) detection algorithms that can be carried out by digital relays or external computation resources.   
Artificial Intelligence (AI) can significantly improve the detection of HIFs by overcoming limitations of traditional protection methods. Since HIFs produce weak, erratic, and nonlinear fault signatures that vary depending on environmental and system conditions, AI is well-suited to recognize subtle patterns in real-time data. AI provides a powerful and flexible approach to detecting high impedance faults by learning from real-world data, extracting complex patterns, and adapting to new conditions. Its integration into modern protection schemes could represent a paradigm shift toward intelligent, data-driven fault detection, enhancing both grid safety and operational reliability.
PG&E and SEL are collaborating with Microsoft to develop advanced AI-driven solutions for wildfire detection and prevention. The initiative focuses on analyzing months of continuously recorded high-sampling electrical waveform data from actual distribution feeders to identify early indicators of wildfire risk - particularly subtle, high-impedance faults that traditional protection systems often miss.
In this presentation, collaborators will share the AI model development learning, emphasizing practical feasibility and commercial viability. We will share how the AI theoretical exploration transcendent into actionable operations insights and mitigate wildfires.