Prognostic Health-Management System Development for Electromechanical Actuators

Abstract

systemthatdiagnoseselectromechanicalactuatorfaultsandemploysprognosticalgorithmstotrackfaultprogression and predict the actuator’s remaining useful life. The diagnostic algorithm is implemented using a combined modelbased and data-driven reasoner. The prognostic algorithm, implemented using Gaussian process regression, estimates the remaining life of the faulted component. The paperalso covers the selection of fault modes for coverage and methods developed for fault injection. Validation experiments were conducted in both laboratory and flight conditions using a flyable electromechanical actuator test stand. The stand allows test actuators to be subjected to realistic environmental and operating conditions while providing the capability to safely inject and monitor propagation of various fault modes. The paper covers both diagnostic and prognostic run-to-failure experiments, conducted in laboratory and flight conditions for several types of faults. The experiments demonstrated robust fault diagnosis on the selected set of component and sensor faults and high-accuracy predictions of failure time in prognostic scenarios.