What Predictive Analytics Look Like In The World’s Largest Industries

As large industrial machines take on massive tasks, they’re not only being challenged like never before, they’re also being judged for more than just productivity. Whether transporting Chicagoans, harvesting soybeans, or mining iron ore, these machines must also talk effectively with their operators and other machines, “learn” as they operate, and become a force for driving business outcomes.

It Starts with Data

At Uptake, we’re at the forefront of machine intelligence. Our analytics pinpoint which of our customers’ parts may fail, while offering guidance on how to prevent those failures. By gathering mountains of data and processing for real-time insights, our customers — and their ever-smarter machines — operate more efficiently.

To do that, our analytics use data from many sources — everything from sensors on machines to handwritten maintenance records. For locomotives at a railroad, for example, analytics might look at engine temperatures, voltages from the electrical system, histories of visits to repair shops, and even weather during the locomotive’s trips.

Like People, Machines Have Norms

With every new piece of data, a complete picture of the machine’s normal operation emerges. Like a doctor who sees a patient for years and gets to know their norms, the software learns what’s typical for a particular machine. Patterns are established, and problems become easily identifiable — before they happen.

Moreover, with so much data trending, the platform evolves continuously. In the end, everyone benefits, as predictions become increasingly precise across multiple industries.

Flagging Health Issues

After recognizing a problem on the horizon, our system issues an alert. With a mining truck for example, Uptake’s technology could tell a mining company that a particular vehicle’s brake system will fail in the next two weeks. From there, the software shifts into troubleshooting mode and uses patterns in data to locate the source of the problem. The software is able to pinpoint exactly where component health is degrading — for example, a failing brake pump.

Showing a customer where to look and what to fix drives operational productivity. When we provide warning about an unexpected and more expensive impending failure, customers can schedule focused, preventative maintenance, maximize uptime, and increase return on an asset.

Feedback Loops Drive Continuous Improvement

Our customers expect their organizations to get better over time. We expect the same from our platform, and that’s why we’ve designed it to learn automatically, both from new data and from feedback. After our platform makes a prediction, diagnosis or recommendation, we make it easy for users to tell us whether that course of action was effective. In fact the system scores its own effectiveness, then incorporates that knowledge to build increasing precision moving forward. After all, the more advance warning we can provide, the more our customers than optimize their assets, systems and fleets.

What’s needed to bring outstanding productivity and efficiency to the world’s largest industries? Useful data, great data scientists, the right models, and industrial leaders who want to innovate and disrupt.

Brian Silva is a Data Scientist at Uptake.