Adaptive Learning for Stall Pre-cursor Identification and General Impending Failure Prediction

Frontier Technology, Inc. (FTI) and Northeastern University propose to investigate and develop an innovative approach to predict stall events of aircraft engines prior to occurrence and in sufficient time to allow the FADEC controller to adjust engine variables. The team will utilize vector quantization and neural network techniques to develop accurate models of engine behavior that will be used to detect and predict the stall. Vector Quantization and transfer function models will be used to create the models that estimate engine current conditions. These conditions and in-situ sensor readings are provided to a Neural Network (NN) to predict the occurrence of a stall. Engine data will be provided from GE Aviation will be sued perform both the vector quantization and to train the NN model. The research team has extensive experience working with engine data to detect and diagnose faults and to predict impact on engine performance. Northeastern University has performed a GE-sponsored project to predict engine stalls and other fault events that is closely related to the proposed technology. This effort extends FTI’s research into engine failure detection and prediction analysis which has been performed in support of the US Navy and US Air Force.