Fault detection and isolation of gas turbine engine using inversion-based and optimal state observers

This paper aims to design, develop and compare fault detection and isolation (FDI) schemes including inversion-based fault reconstruction and optimal state observers for a class of nonlinear system subject to concurrent faults and unknown disturbances that represents the nonlinear dynamic model of a gas turbine engine. Towards this end, for each fault, utilizing an existing coordinate transformation, the orig-inal system is transformed into a new form in which both observers are applied for fault diagnosis. The coordinate transformation comes from observability concept in differential geometry. The inversion-based observer is highly beneficial for straightforward detection and directly isolating the faults, and the state observer is optimal with respect to the magnitude of observer gain and convergence time. The mentioned approaches are implemented for FDI of a gas turbine model subject to compressor efficiency, compressor mass flow capacity, and actuator faults in addition to an unknown disturbance. The simulation results illustrate that the proposed FDI schemes are a promising tool for the gas turbine diagnostics. (c) 2020 European Control Association. Published by Elsevier Ltd. All rights reserved.