AI & Analytics
Army leverages machine learning to predict component failure
- By Matt Leonard
- Jul 02, 2018
The Army will be using machine learning software to predict when components on the Bradley Fighting Vehicle need maintenance.
Through an award facilitated by Defense Innovation Unit Experimental, the Army will be working with Uptake, a company that provides artificial intelligence solutions for industrial sector clients, to predict component failures, decrease the frequency of unscheduled maintenance and improve the productivity of repair operations.
“The Bradley already has sensors throughout it on major components of systems, so what we’ll be doing is capturing that data that is generated all throughout the Bradley and then combining it with our software to provide insights on potential failures before they happen,” said Matthew Lehner, a spokesperson for Uptake.
None of the company’s software is uploaded to the military vehicle. The data from the sensors is transmitted to the cloud, where Uptake's software analyzes normal operational patterns and learns to predict failures.
The company will also leverage the billions of hours of operating data it has collected from the other industries it serves, Lehner said. “We have 230 million hours of data on diesel combustion engines, which is something the Bradley also has,” he said.
The Uptake interface provides insight on individual vehicles. If a fault is found, then it is listed along with a fault code, a description, the severity (low, medium, critical), and the first and last occurrence of that particular fault.
The interface largely stays the same from industry to industry, but the machine learning models have to be changed and require a verification period before full deployment, Lehner said.
In 2016, the Army Materiel Command's Logistics Support Activity worked on a similar project with IBM that used the company's Watson artificial intelligence platform to help predict maintenance problems in Stryker combat vehicles.
Matt Leonard is a former reporter for GCN.