Multiobjective genetic programming for gas turbine engine model identification

Abstract

A genetic programming-NARMAX approach has been successful applied to the identification of nonlinear systems. This evolutionary identification method has been extended to a multiobjective form with the aim of simultaneously optimising different measures of the system model under investigation. So far, this novel approach has only been tested on simulated data. Here, we demonstrate a practical application of this technique to obtain a model of the relationship between the fuel feed and the shaft speed dynamics of a gas turbine engine.