Fault detection and classification for flight control electromechanical actuators

Future aircraft architectures will incorporate more energy-efficient electromechanical actuators (EMA) for flight controls actuation. Development of reliable health monitoring techniques for EMAs promises to maintain or even increase the overall availability and safety of these new aircraft designs. When it comes to EMAs and similar mechanisms, certain fault types clearly manifest themselves through loss of functionality. Other faults, referred to as latent, do not immediately result in a significantly compromised actuator performance, thus making them challenging to detect. This paper presents a new vibration-based hybrid technique for detecting latent EMA faults without needing an initial stage of fault feature learning. The two faults considered in the study are a high-criticality jam and a low-criticality spall (metal flaking) in the actuator ballscrew mechanism. The actuator position is used to resample variable-speed vibration measurements of a single accelerometer into constant-rate measurements. A set of health characterization signatures is derived theoretically based on the EMA ballscrew kinematics. These theoretical signatures are compared with the signatures extracted from vibration signals measured experimentally on the EMA test articles. The vibration signatures approach is also compared to the diagnostic approach based on EMA motor current measurements. The ability to detect and classify latent faults early as high- or low-critical can improve maintenance planning and increase aircraft dispatch reliability. The technique has been validated on fault-injected data sets collected on the NASA Ames Research Center Flyable Electro-Mechanical Actuator (FLEA) test stand