The Simulation and Analysis for Neural Network Adaptive Control of Aero-engine

A neural network adaptive control structure for a certain aero-engine was presented. The problems such as structure, off-line training, determination of initial weight values and control network involved in the control loop were studied. The simulation of the engine control system indicates that aero-engine steady-state control can achieve good performance with this manner. When deviating from the design point, the parameter of the neural network will be modified on-line with its own strong learning and adaptive ability, which make the control system maintain good performance and strong robustness.