Redundancy-Aware Flight-Log Health Scoring for Unmanned Aircraft Propeller-Fault Triage

Redundancy-Aware Flight-Log Health Scoring for UAV Propeller-Fault Triage: A Cross-Airframe Calibration Gap

This study evaluates redundancy-aware Artificial Age Scoring across all 130 DronePropA flights. Nested validation on 120 flights demonstrates interpretable within-airframe propeller-fault triage, while testing on ten healthy flights from two unseen drones identifies an important cross-airframe calibration gap: a model calibrated on the main drone falsely classified eight external healthy flights as faulty. The results show that redundancy adjustment can prevent repeated evidence from inflating a health score, but cannot by itself correct platform-dependent shifts in control-torque and motor-command distributions. Consequently, fleet-level deployment requires airframe-aware calibration and faulty-flight validation on unseen aircraft. Reproducible code and machine-readable results are available at https://github.com/seymus1/DronePropA-RA-AAS