The little bias state space modeling and neural network building for predicating on Aero-engine

The article gives an overview of the linear state space modeling of the Aero-engine; uses the scheme of central difference to solve the partial derivative equation of certain a turbofan engine establishing a little bias state space model. And the simulation between the nonlinear model and the above-mentioned model step responses indicate the state space model is so closed to the nonlinear model. In addition, the article designs a kind of BP neural network under the fixed height (H) and Mach number (Ma) to predicate the values of the little bias state space model parameter. According to the simulation results, the BP network can predicate the bias state model´s parameter accurately along with the changing of the engine fuel flux (Wf). It can used to be a prognosticated model under the flight envelope fixed values of H and Ma.