ABSTRACT
In the last three chapters we focused on Kalman filters, neural networks, and fuzzy logic systems for fault isolation in gas turbines, respectively. Fuzzy systems are also universal function approximations in a manner similar to that in neural networks. Fuzzy systems also address the issue of uncertainty using a built-in fuzzifier, whereas a neural network learns the noise characteristics of the data through training. It was shown in Chapter 10 that fuzzy systems provide accurate fault isolation results for gas turbine diagnostics. However, the neural and fuzzy methods for diagnostics are highly configuration dependent, meaning that if the underlying model used to obtain fault signatures or the measurement uncertainties of the signal changed, the diagnostic systems have to be redeveloped. Since there are many different engines operating with different airlines, there are likely to be many possible combinations of fault signatures and measurement uncertainties for the fault isolation systems that need to be developed. Very often the process of redeveloping the underlying numerics or rules for the diagnostic system is a trial and error process that can be very tedious and requires considerable human effort.
