In connection with the start of Wildfire Awareness Month, Pacific Gas and Electric Company announced its Continuous Monitoring Center, which it’s calling a “first-of-its-kind centralized hub” to “detect risk early, prevent wildfires and strengthen reliability for customers.”
The CMC in San Ramon uses “advanced machine learning” from tens of thousands of sensors across PG&E’s electrical grid, combining that with data from its smart meter network (with 5.5 million of them scanning grid conditions).
“By analyzing this data continuously, trained experts can identify emerging issues—often before customers are impacted—and dispatch crews to take action before those issues escalate into wildfires or outages,” a spokesperson wrote in a May 1 release. “Much like a smart watch alerts you to changes in your health, the Continuous Monitoring Center analyzes signals across PG&E’s electric system and flags abnormalities that require attention.”
PG&E’s machine-learning models will look for patterns pointing to heightened risks, according to company officials.
During tests in 2025, these system surveillance capabilities intercepted 17 potential ignitions in high fire-risk areas, avoided 12 million minutes of unplanned customer outages, reduced emergency outage response time by 2,620 hours and saved around $6 million in operations costs, per PGE data.
“The threat of wildfire requires more than incremental improvement, it demands a different kind of vigilance,” Mark Quinlan, PG&E’s senior vice president of wildfire, emergency and operations, said in the release. “With the Continuous Monitoring Center, we’re adding another layer of protection, using predictive intelligence from millions of data points across our system, to spot problems forming before they become emergencies. The results are clear: faster detection, quicker action, a safer grid and real cost savings for the customers we serve. We are also actively sharing what we’ve learned with utilities and industry peers around the world.”
In one case, the Continuous Monitoring Center team of engineers caught an issue on the Brunswick 1106 circuit in Nevada County.
“The machine learning model identified a possible wiring issue and, after traveling to the location, an electric troubleshooter found melted insulation at the transformer caused by degraded connections because of severe weather-related stress,” a PGE spokesperson wrote in the news release. “Crews replaced the transformer and other associated equipment before an ignition could occur. Had an ignition resulted, it could have led to a 17-acre wildfire, potentially damaging two or three structures and causing up to $1.4 million in losses.”
PG&E has recorded 1,484 of what they’re calling “good catches,” since 2025.
The CMC analyzes data from multiple technologies:
- Early Fault Detection Sensors – Radiofrequency (RF) monitoring that identifies partial discharge, arcing and insulation breakdown, across 900 circuit miles in high fire-risk areas.
- GridScope Devices – Pinpoints the location and nature of issues across 1,350+ circuit miles in high fire-risk areas.
- Downed Conductor Detection – Enables fast identification of the location and cause of disturbances across 1,350+ circuit miles, supporting quicker response and targeted repairs.
- SmartDetect – Uses SmartMeter™ data and machine learning to monitor grid performance, identify patterns and anomalies that may indicate emerging risks.
- Distribution Fault Anticipation Sensors – Identifies small system disturbances across 8,900+ circuit miles that enable proactive maintenance to prevent outages or risks before they occur.
- Line Sensors – Detects changes in electric current patterns that help crews locate issues faster and determine root causes for more effective repairs across 19,000+ circuit miles.
- Grid Data Analytics & Technology Platform – Integrates incoming sensor and outage data across PG&E’s electric grid into a single system to help identify trends, prioritize preventative work and inform faster decision-making.
