Abstract
A neural network approach is employed for estimating key efficiency parameters in a gas turbine engine. The concept is demonstrated within a limited operating region for a given engine. The neural network is developed to estimate certain unmeasurable parameters in a first-principles mathematical model of the engine. The network is trained using data derived from measured data taken on an auxiliary power unit (APU) engine (from an aircraft application). A discussion of the neural network development and its application to on-line fault detection in an industrial gas turbine engine is also presented. The technique could be used for condition-based maintenance or to monitor the energy efficiency of an industrial gas turbine.
Gabel, S. (2003). Using Neural Networks. Energy Systems Laboratory (http://esl.tamu.edu). Available electronically from
https : / /hdl .handle .net /1969 .1 /91032.