Frequency response of gas turbine condition monitoring system

Abstract

The work reported in this thesis is concerned with the Condition Monitoring by using Generalised Frequency Response Function. Two types of nonlinear models are discussed but the attention is focussed on a nonlinear model called NARX (Nonlinear AutoRegressive with exogenous input). The first of two parts of the thesis deals with the theory. Initially, the NARX and NARMAX representations are introduced. Recursive Least Square and Extended Recursive Least Square for estimating the unknown parameters in the NARX and NARMAX models are presented. A discussion of the techniques for selecting the "best" model is given. To complete the modelling techniques, the validity test of the identified model are presented. By using the NARX model, an analytical expression for calculating the n-th order of Generalised Frequency Response Function in term of model parameters and the gain bound can be obtained. The second part of the thesis deals with the case studies with practical identification of nonlinear system using the written package. Six case studies are demonstrated.Techniques for obtaining the optimal model, validity test are described. By using this optimal model, the decomposition of Generalised Frequency Response Function can be visualised. Finally, the Energy Density of the model is presented.