Predictive Maintenance

AI-Enhanced Predictive Maintenance for Aircraft Health Monitoring

Predictive Maintenance

Two digital-twin prototypes that turn raw flight telemetry into ranked maintenance alerts.

The Challenge

Long-range unmanned aerial systems (UAS) require reliable health monitoring to support safe, scalable operation, yet maintenance is often based on fixed schedules rather than the actual condition of the aircraft. This can result in replacing components that still have useful life while missing degradation that begins between scheduled inspections.

Detecting degradation early is particularly challenging when aircraft have different sensor configurations and limited onboard computing resources. A scalable approach was needed to continuously assess subsystem health, identify emerging faults, and provide actionable information for maintenance decisions.

Our Approach

As part of a NASA SBIR Phase I program, Fraunhofer USA CMA collaborated with a leading UAS fleet management platform provider to develop an AI-enhanced stochastic digital twin for continuous health monitoring of UAS.

The platform compares expected subsystem behavior with measured behavior using flight telemetry and historical reliability data. Detected deviations are evaluated and ranked by severity, generating interpretable Advisory, Caution, and Warning alerts rather than raw anomaly scores. The platform-agnostic architecture is designed for compact edge hardware and variable sensor inputs, supporting deployment across different aircraft platforms and future aviation safety frameworks.

The Impact

Two working digital-twin prototypes were demonstrated using real flight telemetry: one for propulsion monitoring on a fixed-wing UAS and another for flight-control monitoring on a rotorcraft UAS. Both prototypes detected and classified simulated degradation and fault scenarios and translated them into severity-ranked maintenance alerts.

The project demonstrates how AI-enhanced digital twins can shift aircraft maintenance from fixed schedules toward condition-based, predictive maintenance, providing a foundation for continuous health monitoring and more proactive maintenance aligned with NASA's future aviation safety framework.

Ready to maintain on condition instead of the calendar? Get in touch.

Contact Press / Media

Dr. Jeno Szep

5700 Rivertech Court, Suite 210
Riverdale, MD, 20737-1250