Early detection of fan-blade faults in aero-engines is a key indi- cator of engine operation and enables effective maintenance planning. The structural health of turbine blades directly affects thrust generation and over- all engine reliability. Hence, continuous monitoring of blade condition crit- ical for safe aircraft operation. However, assessing rotor imbalance through vibration signatures typically requires extensive instrumentation, multiple accelerometers, and rigorous calibration procedures, thereby increasing test effort and integration complexity. This study presents a hybrid, AI-enabled virtual sensing framework for aero-engine condition monitoring in con- trolled engine test environments. In place of deploying numerous piezoelec- tric accelerometers, the framework employs an XGBoost regression model integrated with order-tracked vibration analysis to estimate vibration re- sponses at locations where physical sensors are not installed. This approach reduces wiring complexity, minimizes instrumentation burden, and im- proves maintainability. Two physical accelerometers are substituted with XGBoost-based virtual sensors trained using reference vibration measure- ments and key operating parameters, including Fan Rotor (N₁) shaft speed and Power Lever Angle (PLA). FFT-based order tracking is utilized to ex- tract shaft-synchronous components (1× and 2×), which serve as sensitive indicators of blade structural anomalies and Foreign Object Damage (FOD). These order components are strongly correlated with variations in mass im- balance and blade deformation, enabling reliable detection of early-stage faults. Validation across multiple engine test cycles demonstrates strong agreement between predicted and measured vibration signals, yielding cor- relation coefficients exceeding 0.93. The proposed physics-informed virtual sensing approach provides real-time monitoring capability, enhances diag- nostic coverage, and significantly reduces instrumentation wiring and test preparation time required for aircraft integration. This framework contrib- utes to improved aero-engine safety, reduced test effort, and more efficient condition-based maintenance practices.
Srinivasan et al. (Tue,) studied this question.