GE Digital believes that edge computing will play a key role in the industrial Internet of Things (IoT), and that’s why the company announced last October that it was expanding its Predix industrial IoT platform’s capabilities to better handle computing at the network’s edge. But how exactly is the company accomplishing this?

In an interview with SDxCentral, GE Digital’s Gytis Barzdukas, vice president and head of product management for Predix, said that the company is adding analytics and machine learning to its Predix Edge Manager, which serves as a portal to provision and manage devices. The analytics are then combined with machine learning (ML) concepts so the company can help customers predict certain behaviors.

Barzdukas used Schindler Elevators as one example. The company has thousands of devices that it manages by using the Predix Edge Manager. Those devices are used to monitor the behavior of banks of elevators. And by incorporating machine learning and predictive analytics, the Edge Manager can determine if an elevator door is sticking and will then notify Schindler that the elevator needs maintenance.

The machine learning technology that the Predix Edge Manager incorporates comes from the company’s 2015 acquisition of Wise.io. Barzdukas said the company started testing the Wise.io machine learning technology last year to see how it performed in certain use cases. “We want to train models in the cloud to pull that unique data off of the edge and put it in the cloud so you can do more algorithm training,” he said. The goal, Barzdukas added, is to use new data gathered at the network edge to improve the predictions. “The data is then used to train the model and then push that information down to the client,” he said.

GE Digital is also incorporating the machine learning capabilities into work its doing with the gas industry. The company is working with pipeline inspection gadgets that are sent through a pipeline to take pictures in search of cracks. Barzdukas said that currently there is a very manual process for checking those photos where dozens of analysts comb through the images.

Instead, GE Digital is creating a way to take the data and the photos and use algorithms and machine learning to predict where faults in the pipeline may occur. “We can apply the Wise.io ML technology to train the machines to have predictive capabilities,” he said.

But Barzdukas admitted there are challenges to growing the business. Because each industry needs specific algorithms and data, it can be difficult to scale and create more solutions for other industries. “Hiring data scientists is tough today,” he said. “We need to figure out how to scale our business.”

Nevertheless, the Predix platform is growing. In October, GE Digital said that Predix has more than 1,000 customers and more than 960 ecosystem partners. The platform also now operates in more than 85 countries.

Copyright: 123RF Stock Photo