PHI-Twin: a digital twin that watches landing gear the way a mechanic would, at machine speed.

Most landing gear health monitoring still leans on scheduled teardown inspection — safe, but blind to the time between inspections. PHI-Twin runs continuously instead, fusing multi-channel sensor data through a physics-informed neural architecture that's learned what normal wear looks like versus what a developing fault looks like, and flags the difference in milliseconds, at the edge, without a round-trip to the cloud.

We validated the approach on a Dornier 228-212 airframe model across eleven distinct fault classes, and the architecture is built to generalize to other gear-equipped platforms.

Full model architecture, training methodology, and dataset composition are reserved for OEM and institutional partners under NDA — this page covers what the system does, not how it's built.