On Line Fault Detection and Isolation in Gas Turbine Combustion Chambers
This work presents the development of a new model for the exhaust gas temperature based on both basis function expansion and the Brayton cycle. This model is a function of the power generated, ambient temperature, compression rate, and the temperature of each combustion chamber. This last temperature is unknown, but could be estimated. The model basis functions also include the spatial distribution of the combustion chamber and exhaust gases swirl angle surface. Thus, based on the gas path in the turbine, each base function of the model is related to a particular combustion chamber. This is the main assumption that allows solving the fault detection and isolation problem in gas turbines at the level of combustion chambers. As a result of the model identification at every instant, there is a group of coefficients, which are associated to each combustion chamber. From these coefficients, it is possible to generate signals that can be analyzed with statistical techniques and also with wavelets to detect abrupt changes in its behavior.