In order to control the turbojet engine in the whole flight envelope, this paper will establish the self-adaptive PID neural network control on the basis of the combination of the identification network of the RBF neural network and the controller of the BP neural network. RBF neural network adopts the offline training and the on-line adaptation of weight and bias. To speed the convergence, it will present a detailed discussion on adopting the gradient descent method with special inertia item. Massive stimulation prove the superiority of the RBF network to the ELM and the standard and the improved BP. Stimulations to the pilotless aircraft turbojet engine prove several advantages of this controlling method such as strong robustness, swift response, and minimal steady-stable error