Neurofuzzy Identification and Control of a Gas Turbine Jet Engine

In this paper the neurofuzzy network identification and control of an aero gas turbine jet engine is described. The system dynamics belong to a specific class of nonlinear systems whose linear parameters are unknown nonlinear functions of some measurable operating points of the system. A neurofuzzy unit is used to identify each nonlinear function; each unit contains a set of linguistic rules whose inputs are the measurable operating points of the system over the operating space and whose outputs are estimated values of the nonlinear parameters. The estimated values are then used in the design of controllers at critical points in the flight envelope which are then synthesised by gain scheduling against the operating point. Convergence and stability results for the modelling algorithm can be proven.