IoT Evolution followed the cycle of the Edge at GE as we were guided through the company’s Predix Cloud offerings at the edge during the recent GE Edge Symposium, at the Global Research Center campus outside Albany, New York.
Wes Skeffington, Senior Principal Engineer, Edge Controls, was our guide through the tour at the GE Edge. Skeffington first brought us to the Field Agent Station, which is the new edge processing module at the heart of many of GE’s edge implementations. It handles security and can manage control functions through a variety of apps which can be installed and linked to the cloud.
The next stop was at the Mini Field Agent station, which is used mostly for asset performance management (APM), and several can be linked to a full Field Agent and share data sets to create an asset model. This process takes the data from the edge and brings it into the APM for the purpose of Identity Management and, when married to data analytics both at the edge and in the Predix cloud.
In a practical example, the Mini and full Field Agents are connected to power generation turbines to gather and process data, which is fed into the Predix Cloud and used to build what GE calls a Digital Twin.
The Digital Twin is the most interesting piece of the entire equation, to me. It is used to allow the analytics engines and human data scientists to visualize and estimate problems based upon real, live data. And that allows the digital twin to solve one of the most pressing problems I’ve had with edge processing since I first started hearing about it. That problem was, in short, that analytics performed at the edge cannot have a global sense of statistical data, and therefore the results are inherently less trustworthy, especially with respect to predictions of future behavior. And that, to me, meant that it wasn’t viable for mission-critical applications.
Well, with the Digital Twin, one digital twin can be created for each turbine, and another for all turbines (or other equipment) of a like model type. That global Digital Twin would be built with inputs from all the Field Agents connected to those devices and can therefore be built and fed with global statistical data. This all allows the “local” Digital Twin to be useful for predicting local phenomena and a “global” Digital Twin can be used for more meta data analysis and predictive maintenance.
And that’s a huge step forward in the viability of Edge processing for IIoT, if you ask me.
Many thanks for the demos from the GE experts who explained all this to us:
Kyle Hable, GE Automation & Controls
Ravi Subramanyan, GE Digital
Antonio Mejia, GE Aviation
Scott Kophcho & Tim Jansen, GE Power
Wes Skeffington, GE Global Research
Paul Caffrey, GE Digital Power
Ken Briodagh is a writer and editor with more than a decade of experience under his belt. He is in love with technology and if he had his druthers would beta test everything from shoe phones to flying cars.
Edited by
Ken Briodagh
