159515752 | Everythingpossible | Dreamstime
Dreamstime M 159515752

Machine learning helps IT, OT teams anticipate equipment breakdowns long before they happen

Oct. 5, 2023
Two Seeq senior analytics engineers show SI webinar attendees how ML helps build “anomaly-detection” models that plant managers can use to manipulate their data, anticipate possible downtime events, and engage preventive maintenance.

Downtime is anathema to any strong and streamlined manufacturing operation. Machines still break down, so some downtime is unavoidable, but what if OT and IT personnel had the toolsvirtual Deloreans driven by Doc Brown and Marty McFlyto peek into the future and anticipate breakdowns in time to head them off with preventive maintenance.

Seeq senior analytics engineers Emilio Conde and Sean Tropsa were on the scene Oct. 5 to show off the Seattle-based company’s advanced analytics tools for processing the tons of data that manufacturers gather from their operations. Thursday’s webinar, which was packed with attendees, was sponsored by Seeq as part of this week's Smart Industry Fall Insight Series, which concludes on Oct. 6. Editor-in-Chief Robert Schoenberger moderated the engaging hourlong chat. A recording of the event is available now.

Next up: How To Capitalize On Generative AI In Your Manufacturing Business

AI and machine learning play prominent parts in the future-predicting magic that Seeq’s analytics SaaS solutions wield, and Conde and Tropsa demonstrated some scenarios where Seeq’s solutions could prevent internal breakdowns in plant machinery that can cut into the bottom line.

They spoke a lot on Thursday about "process optimization" and “anomaly detection” models, and a slide in their talk presented the example of system bearings, where the failure of one could lead to the failure of more, causing downtime that can be very costly for an industrial operation.

Seeq’s solution? SMEs cleanse available data on the machine’s operation, they identify normal operation, calculate statistical thresholds—in this case the maximum stress on bearings and the point where they might fail—and identify deviations from their normal performance.

The result? Maintenance is notified of abnormal operation days before a possible failure. That's the definition of predictive maintenance.

See also: Surf’s up! Wave-pool designers detail how they use digital twin and control systems tech to simulate the sea

Seeq’s solutions aren’t one-size-fits-all, Tropsa noted. The company has preview servers available for any industry IT personnel who want to test Seeq's solutions and see how they might be adapted to fit their employers' operations.

“There is no one algorithm that will solve all problems,” he added. “This is why it’s important to have flexible analytics platforms.”

And a plant process engineer doesn't necessarily need to know complex software coding to use Seeq's solutions because they were originally designed for process engineers, who might not necessarily hold that knowledge, though taking advantage of some capabilities since the original release make coding acumen handy.

Seeq's solutions are all about manufacturers reaching their their digital transformation goals. More than 88% of companies fall short.

"Assess how your digital transformation is going," Conde advised. "Many organizations are data rich, evaluation poor."

About the Author

Scott Achelpohl

I've come to Smart Industry after stints in business-to-business journalism covering U.S. trucking and transportation for FleetOwner, a sister website and magazine of SI’s at Endeavor Business Media, and branches of the U.S. military for Navy League of the United States. I'm a graduate of the University of Kansas and the William Allen White School of Journalism with many years of media experience inside and outside B2B journalism. I'm a wordsmith by nature, and I edit Smart Industry and report and write all kinds of news and interactive media on the digital transformation of manufacturing.