Application of Fuzzy Logic for Fault Isolation of Jet Engines

Abstract

A fuzzy logic system is developed for gas turbine module fault isolation. Inputs to the fuzzy logic system are measurement deviations of gas path parameters from a “good” baseline engine. The gas path measurements used are exhaust gas temperature, low and high rotor speed and fuel flow. These sensor measurements are available on most jet engines. The fuzzy logic system uses rules developed from performance influence coefficients to isolate the module fault while accounting for uncertainty in the gas path measurements. Tests with simulated data show the fuzzy system isolates module faults with accuracy of over 95 percent. In addition, the fuzzy logic system shows good performance even with poor quality data. Additional pressure and temperature measurements between the compressor and before the burner help to increase the accuracy of fault isolation at high levels of uncertainty and when modeling assumptions weaken.