Diagnosis of turbine engine transient performance with model-based parameter estimation techniques

This review surveys knowledge needed to develop an improved method of modelling the dynamics of gas turbine performance for fault diagnosis applications. Aerothermodynamic and control models of gas turbine processes are examined as complementary to models derived directly from test data. Extensive, often proprietary data are required for physical models of components, while system identification (SI) methods need data from speciallydesigned tests. Current methods are limited in: tuning models to test data, non-linear effects, component descriptions in SI models, robustness to noise, and inclusion of control systems and actuators. Conclusions are drawn that SI models could be formulated, with parameters which describe explicitly the functions of key engine components, to offer improved diagnostic capabilities