Over the past few months, GE Intelligent Platforms has made several announcements in the manufacturing and processing space, ranging from MES/PLM connections to mobile HMIs. Now the company is announcing what it says is the first industrial application of big data management and analytics.
The Proficy Monitoring & Analysis suite contains six integrated products (four existing and two new products) that can also be purchased separately. The four existing products in the suite are: Proficy Historian data collection software; Proficy Historian Analysis for data mining and visualization; Proficy SmartSignal for predictive analytics for condition-based monitoring; and Proficy CSense to troubleshoot process problems, monitor process health and create closed loop process optimization. The two new products are: Proficy Historian HD, which allows storage of massively large data sets in a Hadoop cluster (see video at bottom of this article for a basic Hadoop tutorial); and Proficy Knowledge Center, a console that connects the suite with process visibility, asset health assessment and process optimization.
Brian Courtney, general manager for GE Intelligent Platforms Industrial Data Intelligence Software group, says the target user for this software suite is “anyone monitoring fleets of assets to get efficiencies” across those assets. He adds that companies in the oil and gas, power generation and mining industries are among the first to begin adopting this type of software.
Explaining the use of Hadoop in an industrial application, Courtney provides an example from GE Aviation’s experience with the software. “Consider seeing an unusual start up sequence in a turbine and wanting to know if you’ve ever seen that sequence before. Historians aren't good at this type of signal comparison. For a turbine you'd have to look for all single starts and pull up a five-minute window for every start and compare them all. In the end, you’ll get back so much data you’ll run out of virtual memory. But we’ve been able to do this in a Hadoop cluster and get back results in less than two minutes on a 40 terabyte data set. The results showed that we had seen this kind of unusual start up 12 times over the last 10 years across 1,700 assets. We were able to use that information to determine the time to failure for that turbine. The ability to analyze data this way enables a new type of analytic.”
Hadoop is based on technology developed for Google, Yahoo and Facebook for tracking the world’s clickstream data, says Courtney. Hadoop is a “massively parallel processing system that’s mostly been used with time-stamped data to optimize targeted advertising,” he says. “This technology has not been applied in the industrial sector until now. The issue in adapting it to industrial use has been in adjusting it to deal not only with time-stamped data (something which actually occurred), but also to handle time series data (estimation based on measurements taken over a series of time). The underpinnings of Hadoop had to be adapted to understand time series data.”
GE claims that Historian HD, using Hadoop, enables users to bring together an entire set of historical time-series data while reducing data storage costs by as much as 85 percent.
“We use the software ourselves in our own monitoring and diagnostics centers to manage trillions of dollars in asset value,” Courtney says. “Today, in the GE Industrial Performance and Reliabilty Center, GE engineers monitor thousands of mission critical assets for our customers to ensure uptime, asset reliability overall production throughput."
Proficy Monitoring & Analysis suite is available via license or as a service.
The video below is a general tutorial on Hadoop.