Application of neural network approximation methods in the generalization and presentation of the aircraft gas turbine engine turbomachinery characteristics

Abstract

A technique of generalizing and representing the characteristics of gas turbine engine (GTE) compressor and turbine using the neural network approximation methods is given. The algorithm of generalizing the experimental performance in mathematical packages on methods of artificial neural networks is shown. Results of approximating and interpolating the characteristics with the use of neural networks based on a multilayer perceptron with different estimated values of π c * and π t * are compared.