Fault diagnosis for a turbine engine

Fault detection and diagnosis for jet engines is complicated by the presence of engine-to-engine manufacturing differences and engine deterioration during normal operation, the complexity of an accurate engine model, and our inability to directly measure certain engine variables. Here, we work with a sophisticated component level model (CLM) simulation of a turbine engine (XTE46) that can simulate the effects of manufacturing and deterioration differences, in addition to a variety of failures. To develop a fault diagnosis system we begin by using the CLM to generate data that is used by a Levenberg-Marquardt method to train a Takagi-Sugeno fuzzy system to represent the engine. The multiple copies of this nonlinear model, each representing a different failure, are then used to generate error residuals by comparing them to the engine output. In fact, we manage the composition of the set of models with a "supervisor" that ensures the appropriate models are on-line, and that processes the error residuals to detect and identify faults. The robustness of the approach is analyzed and several simulations are conducted to illustrate the effectiveness of the method.