Adaptive Turbine Engine Control for Stall Threat Identification and Avoidance

Aurora Flight Sciences and MIT propose to develop a model-based adaptive health estimation and real-time proactive control to identify gas turbine engine stability risks and avoid them through control action. In this concept, the engine control system actively monitors sensors and actuators, compares them against physical models, and infers which components may be performing poorly and may need to be replaced. This estimation procedure will be based on Hidden Markov Model of the deterioration process. In addition, once a threat is identified, the control is given the ability to modify its behavior to avoid operations that increase the chance of compression system instability. Aurora intends to accomplish this through the development of model-based control based on the concept of Rapidly-Expanding Random Trees (RRTs). Using these techniques, maintenance personnel will have the ability to upload health tracking parameters from the control system, directly identifying component performance issues needing corrective action. The engine control system will also be able to adapt to changing deterioration conditions to ensure stall-free operation.