The Problem This Solution Solves
The DoD operates fleets of aircraft from hundreds of global bases and requires year-round maintenance to ensure vehicle readiness and availability. However, maintenance and repair activities are often determined by time-based schedules rather than the condition of the aircraft, which can increase both the rate and duration of unscheduled maintenance due to unforeseen failures and a lack of appropriate parts. Machine learning (ML) solutions that leverage operational performance and maintenance records can anticipate system failure and automate inventory management. These condition-based predictive models reduce the frequency and duration of unscheduled maintenance, increasing aircraft availability by proactively scheduling repairs and decreasing the number of missions aborted because of maintenance-related breakdowns.
In 2019, DIU transitioned a commercial predictive maintenance solution for aircraft, developed by C3.ai, to Air Force and Army customers. Since adopting the predictive maintenance solution, the Air Force has observed a 40% reduction in unscheduled maintenance on high-priority aircraft subsystems, and the Army, in addition to reductions in unscheduled maintenance, has recorded a 24% reduction in severe in-flight malfunctions. In 2020, DIU worked with the Air Force and C3.ai to scale predictive maintenance to more platforms, adding the HH-60 Pave Hawk and the F-35 Lightning II. DIU continues to work with other Services to make predictive maintenance available to more fleets of aircraft.
Separately, DIU is also partnered with the Marine Corps and another artificial intelligence company to prototype and test a commercial predictive maintenance solution for ground vehicles.