‘security Is Non-negotiable’ – Inside The Electric …

As utilities face increasing pressure from extreme weather, aging infrastructure, and the need for faster restoration, automation has become essential.

AI assists control room operators with predictive analytics and situational awareness. Private LTE networks provide the secure, low-latency connectivity essential for coordinating thousands of edge devices and transmitting critical data from inspection sensors to control centers. Together, these technologies can enable utilities to move from reactive repair to proactive, automated response.

A recent panel discussion at DTECH hosted by Factor This titled The Path to Self-Healing Grids: Innovation from Edge to Control Room brought together leaders deploying the three critical layers of a self-healing grid: AI-powered asset intelligence and operations, utility-owned private networks, and grid automation. The panel, moderated by Farnaz Amin, Head of Business Development, Grid Modernization (Americas) – Amazon Web Services, featured Claudia Cosoreanu, Chief Technology Officer, Grid Automation, GE Vernova; Chris Guttman-McCabe, Chief Regulatory and Communications Officer, Anterix; and Emma Stewart, Chief Power Grid Scientist, Idaho National laboratory.

These panelists demonstrated how AI supports operators in making faster, better decisions, AI-driven asset inspection and wildfire risk detection feeding automated prevention, why utility-owned networks are critical for reliable, secure grid automation, integration of edge intelligence with centralized control systems, and next steps toward fully autonomous grid management.

Register now and check out the on-demand webcast for free.

‘Coverage is non-negotiable. Security is non-negotiable’ – Why private networks matter

Why can’t utilities just use available commercial networks for their operations? They work well enough for millions of business use cases – but normal network annoyances and downtime for regular customers can become operational disasters for utilities.

“In our mobile world, we have become accustomed to outages,” Guttman-McCabe said. “We have been, become accustomed to not having coverage. From time to time we wake up and there’s a security breach, and we assume our world isn’t gonna come to an end. All of those things are completely and totally unacceptable to our grid and to our utility operators, right? Coverage is non-negotiable. Security is non-negotiable.”

Another factor utilities must consider is the limitation of commercial networks when attempting to communicate with isolated assets that fall outside of typical coverage zones. Critical operations, wildfire prevention, are full of assets these, Guttman-McCabe argued.

“Some of the most vulnerable wildfire areas have zero wireless coverage,” Guttman-McCabe said. “So if you’re thinking of putting sensors and meters and devices out to, to areas, and you have zero wireless carrier coverage, it’s not an alternative. And so what we’re seeing is dozens upon dozens of use cases coming to us with the idea that ‘we need coverage first, we need security first, we need to be able to prioritize our bits first.’”

AI, edge intelligence, and data sharing

AI is moving the grid from reactive fault detection to predictive threat detection and real‑time decision-making. With edge intelligence, utilities can begin to predict grid events before they happen based on mountains of data. But as more and more data points are added to the mix, making sense of it all is becoming a daunting task.

“There’s just too much right now for humans to even know what’s happening,” Stewart said.

That utility data is important, but not everyone is willing to . Access to data is becoming somewhat of a bottleneck when working with utilities, Cosoreanu argued, as AI requires huge, high quality datasets, but utilities vary in their openness.

“We’re engaging with many customers – Some of them are open to sharing the data, some of them are not,” Cosoreanu said. “Obviously we collect a lot of data with our devices, but the data does belong to the utilities. We’re seeing a few scenarios where the utilities are really interested in keeping the data just for themselves, and they’re building teams to run analytics on that data. And then there’s utilities that are give us some data, not all of it, and we are able to anonymize it.”

Automated systems need to be secure by design to function safely in a utility environment – but they also should be able to determine whether anomalies are simple operations issues or an actual cyberattack. A system that automatically responds, but does so incorrectly, isn’t helping anyone at the end of the day.

“Self-healing is meant to imply that we are able to sort of recover or respond automatically to these items without necessarily a human making all of the decisions, one after the other,” Stewart said. “But we still need to be able to make the right decisions, and we still need to leave the automation in a way that it makes sense and doesn’t cause more problems. There can be challenges as well on the security side of actually understanding where the messages are going. So if you want it to do something, and it’s automated to do that action and then it does something else, we need to know why it did that, and then how to recover from that as well.”

Where can utilities start?

So where can power utilities start on their self-healing grid journey? Many utilities are already using FLISR (fault location, isolation, restoration) to automate fault response and improve outages. Additional tools include high-speed falling conductor protection, which can detect a falling line and de-energize it before it hits the ground. But these deployments have not yet resulted in a truly self-healing grid.

“What we’ve done traditionally with FLISR is not getting us to the self-healing grids,” Cosoreanu said. “And the reason it’s not is it is a pre-designed configuration. So the software will know what to do based on what was designed. Where we need to go to, when we trust the technology, is to allow the system to gather information real-time, design these reconfigurations real-time, and then decide in the most optimal one.”

The panelists agreed that a stepwise but coordinated approach is needed when designing a self-healing grid from scratch: digitize, build communications, deploy analytics, introduce AI, and integrate stakeholder teams early.

“I think getting to a self-healing grid really needs to be step by step, starting with sensors on the grid,” Cosoreanu said. “You have to have visibility in order to let the grid run on its own. And before you get to the point where you let the grid run on its own, there’s steps to take you there.”

That first step, Cosoreanu argues, is digitizing the grid – deploying sensors so utilities can understand in real-time what is happening on the grid. Next is equipment, such as the dynamic system rating offering from GE Vernova. A software layer is then needed to capture all of the collected information and to run the analytics – this software doesn’t need to be AI-driven at the beginning, as teams will need time to get comfortable with the performance of that technology.

A final takeaway? With so much data involved, don’t wait too long to think about cybersecurity, Stewart argued.

“You really need to start doing the cybersecurity or even just security pieces at the very beginning, not at the very end,” Stewart said.

Register here to view the entire on-demand webcast!