Abstract: An aero engine optimal control scheme is presented,which is suitable for all points in the engine whole flight envelope with a neural network approximator.Firstly,local controllers are designed according to local linear models around some design points.Then a neural network approximator is trained to imitate the relation between the parameters of local optimal controllers and the flight conditions and the states of the engine.Finally,a global nonlinear controller is obtained with this approximator.In this approach,nonlinear control system can be designed by the use of many mature linear control methods.And the control system has simple structure,is easy of real time realization.The simulation results show that the aero engine control system has excellent characteristics in both design and non design points in whole flight envelope.