This paper proposes an approach to detecting bearing faults in electromechanical actuators (EMAs) using features extracted from experimental data. The method of feature extraction proposed uses established parameter estimation techniques based on system identification followed by an orthogonal transformation of estimated parameters to derive the required features. A Bayesian classifier is then used to create health classes from the extracted features. The performance of the approach is tested using both data obtained from simulations of bearing faults in a permanent magnet DC motor system as well as data recorded from a Moog MaxForce EMA. The approach shows a misclassification performance of 10% when tested with 50 different data sets generated via simulations. Marginally inferior performance is observed when using 40 different data sets collected from the Moog MaxForce EMA. The conclusion is that bearing fault detection in EMAs is possible via the proposed approach, although further refinements are required.
Nomenclature f cBearing fault frequency reflected in stator current y(t), u(t) Generic discrete-time output and input data e(t) Generic discrete-time white-noise process/Equation-error q −1 , φ Backward shift operator and regression vector v m , i m , ω m Armature voltage, current and angular speed L, T ex Armature inductance and external torque load J, BNet system inertia and net viscous friction coefficient K e , K t
A novel wind machine is described in which an airfoil-like kite is tethered to a moving base; the non-conservative nature of the wind loads over a cycle of the kite allows energy to be extracted at the base. A dynamic model for the kite-tether system is derived which models the tether as a series of point masses connected by massless rods. The lift and drag characteristics of the kite are determined from airfoil calculations. The system has been found to be stable via simulation at several wind speeds of interest. A sample base motion which extracts energy from the oncoming wind is also presented.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.