GE Research Developing AI to Allow Safe Operation of Critical Energy Infrastructure Through a Cyber-Attack
GE Engineers Recently Achieved a Major Technical Milestone With Digital Ghost, Successfully Detecting, Locating and Neutralizing a Cyber-Attack in a Real, Operating Gas Turbine on the Test Stand at GE Power’s Manufacturing Facility in Greenville, Sc.
The Digital Ghost team at GE Research has been awarded a $5.2 million project from the U.S. Department of Energy’s Office of Cybersecurity, Energy Security, and Emergency Response to develop new cyber protection technologies intended to allow critical energy infrastructure like natural gas pipelines to operate safely during a cyber-attack. This capability is at the core of GE Research’s Digital Ghost cyber-physical defense solution and aligned with advanced technology developments the DOE is supporting to safeguard America’s energy infrastructure.
GE’s Digital Ghost represents a new paradigm in protecting industrial assets and systems from malicious cyber-attacks. Originated and developed at GE’s Research Lab, Digital Ghost provides an additional layer of protection beyond the traditional IT/OT firewalls by combining AI and machine learning technologies with sensing and controls to rapidly detect, locate and neutralize cyber-attacks.
“As cyber threats against critical infrastructure increase in frequency and sophistication, we need to be smarter and better at how we use machine learning, controls technologies, and deep knowledge of the process physics to protect these assets and thwart would-be attackers,” said Matt Nielsen, Principal Scientist and project leader. “On this project, we will use the physics of a natural gas pipeline and combine it with machine learning to protect these assets.”
Nielsen explained that being able to map and understand the physics of complex systems such as a natural gas pipeline will allow operators to more quickly detect when and where a cyber-attack may be occurring and how to thwart it. To do this, the project team will create a Digital Twin, or digital model, of a normal operating pipeline and compare that to the actual physical operations of the system. When some aspect of the physical operations fall outside the normal operating mode, advanced algorithms using machine learning and controls will be used to learn where the anomaly is occurring and then act to restore normal operations.
Nielsen said the project will involve the development of a Resilient Energy Delivery and Control System (REDCS) for a natural gas pipeline. “A key capability we’re looking for is the ability to allow natural gas infrastructure to operate safely, even while experiencing a cyber-attack,” Nielsen said. “Working with Intel, Florida State University, and Baker Hughes, we will have the technical expertise and resources we need to both develop and field test our new cyber protection capabilities.”
The three-year project with the DOE is expected to officially kick-off early next year. The goal is to develop the cyber-physical protection technology and then perform a validation field trial towards the end of the project.
GE’s Digital Ghost Technology Achieves Major Technical Milestone
As this new project with the DOE gets underway, GE Research’s Digital Ghost team has been working closely with the GE Gas Power controls team on a separate project supported by the DOE to complete a successful demonstration of Digital Ghost with simulated cyber-attacks in a real, operating gas turbine on the test stand at GE Gas Power’s facility in Greenville, South Carolina. The team was able to inject small, stealthy cyberattacks, which Digital Ghost was able to successfully detect, localize, and then neutralize. The test in Greenville follows successful simulation and lab testing, in which Digital Ghost was been shown to rapidly detect and isolate threats with 99% accuracy.