Prognostics and Health Management of Aircraft Engine EMA Systems

Abstract

In this study, the authors conducted a model-based, engine system analysis of Electro-Mechanical Actuators (EMAs). This effort employed an existing, NASA developed, aircraft engine model. A critical engine actuator within the model was replaced by a dynamic, physics based EMA model that includes: controller, motor, drivetrain and feedback sensor sub-models. The actuator model includes simulation of the electrical, mechanical and thermal response of the system. The resulting platform was used to simulate a range of critical actuator fault conditions including: feedback resolver fault, ball-screw degradation, motor winding short, and LVDT non-linearity. Since the available experimental data from propulsion system EMAs is very limited, this platform provides an ideal opportunity to evaluate and enhance prognostic capability for critical engine applications. The model fault tests were used to demonstrate a prototype prognostics and health management (PHM) system for engine EMAs. First, the system response was used to develop an appropriate mode detection algorithm to identify the ideal system conditions for collection of diagnostic evidence. Then, using the acquired transient and steady-state system response, diagnostic data features were derived from EMA related sensors and engine performance parameters. Using these features as a starting point, a system level reasoner was created using multiple classification techniques including LDA, QDA and SVM. Using model generated data with simulated system variance, it was demonstrated that the reasoner provides excellent fault detection, isolation and severity assessment capability for all considered fault modes. Finally, a suitable actuator life model was developed and a probabilistic prognostic approach was used to determine the remaining useful life of the system. The demonstrated PHM system will significantly enhance the ability to safely operate aircraft, schedule maintenance activities, optimize operational life cycles, and reduce support costs.