New NYU Tandon project to make NYC underground electricity network visible and even more reliable

An advanced data analytics system, assisted by state-of-the-art artificial intelligence (AI), is being developed at NYU Tandon School of Engineering to enhance electric reliability in large cities by detecting grid problems before they cause customers to lose service.

The "Reliability-Enabled Secondary Distribution Visibility" project could transform how utilities monitor the complex, vast meshed network of underground cables and transformers that deliver electricity to homes and businesses. NYU Tandon will conduct initial testing with Con Edison in New York City, home to North America's largest urban underground power grid.

“Detecting and localizing incipient defects in underground power distribution networks is very challenging, as they are not always detectable by humans nor fully identifiable by signals from a single sensor," said Yuzhang Lin, assistant professor in the NYU Tandon Department of Electrical and Computer Engineering and the project's leader. "Our innovation combines AI with power grid physics to analyze data from a variety of sensors across the grid, finding nuances that reveal issues before they cause outages.”

The U.S. Department of Energy's (DOE’s) Office of Electricity has awarded the project's research team $1 million to develop the system with Con Edison, whose service is already the most reliable in the nation. Con Edison will provide testing data from its 43,000 underground transformers and 3.7 million smart meters. NYU Tandon and Con Edison will contribute an additional $400,000 to the research. If successful, the technology could help Con Edison pinpoint grid defects more agilely, potentially making the grid even more reliable.

Itron, which supplies smart meter systems to most major U.S. cities’ utilities, will advise on making the tools compatible with existing utility infrastructure.

At the heart of the project are three key technologies: First, an AI tool that creates accurate maps of the underground power grid. It analyzes data from thousands of smart meters - digital devices installed at homes and businesses that automatically measure and report electricity usage every 15 minutes, replacing old-style meters that required manual readings.

Like a doctor comparing X-rays to textbook images, the AI compares the electricity usage patterns from these meters with utilities' existing models of how power should flow through the grid. This helps spot places where maps may need to be updated to reflect field conditions.

Second, a detection system that uses physics-based statistical analysis of smart meter and transformer sensor data to identify subtle changes in power flow patterns. These changes may be caused by problems like cable burnouts or high-impedance faults — where cables are damaged but still partially conducting electricity. In normal networks with radial structures, such issues would cause immediate outages, but urban grids have myriad redundancies, meaning these conditions can persist unnoticed. The underground nature of urban grids adds to the difficulty of identifying these problems. The sensor data analytics, however, will make hidden defects and hazards visible.

Third, a high-precision, AI-enabled voltage-monitoring system that combines data from smart meters at 15-minute reporting intervals with scattered high-speed sensors reporting every few seconds, to create a detailed picture of power quality across the system. Using deep learning, the system can "fill in the gaps" between meter readings to spot brief voltage fluctuations caused by solar panels and electric vehicle charging, fluctuations that current monitoring systems miss.

The testing ground for this technology will be several areas of Con Edison's vast system, which handles 700 megawatts of distributed solar power and 12,000 EV charging plugs.

If successful, the technology could be deployed in other major cities like Chicago, Washington D.C. and San Francisco that use similar power distribution systems. The technology will be compatible with Itron's widely used smart meter management systems, making it easier for other utilities to adopt.

This DOE award follows another NYU Tandon power grid security project funded last year — DISCOVER — which develops technology to protect against cyber attacks, also in partnership with Con Edison.