基于神经网络逆控制的发动机直接推力控制

首次将动态神经网络逆控制用于航空发动机直接推力控制。为了有效消除由于神经网络逆模型构造误差(即神经网络逆模型不可能完全逼近航空发动机的逆模型)而产生的稳态误差和解决航空发动机推力不易测量的困难,分别设计了积分补偿器和推力估计器,从而实现航空发动机直接推力控制。飞行包线内数字仿真结果表明,此控制方案具有良好的动静态性能、精度高、跟踪快。

Aeroengine direct thrust control based on neural network inverse control

Dynamic neural network inverse control is applied to aeroengine direct thrust control for the first time.For avoiding steady state error caused by construction error of neural network inverse model,which can not be exactly the same as inverse model of aeroengine,an integral compensator which parallelly connects with above neural network inverse model is designed.Dynamic neural network inverse controller is obtained.An aeroengine thrust estimator is designed with neural network for overcoming a difficult problem that aeroengine thrust can not be measured accurately by sensor.Digital simulation results show that this control scheme has good dynamic and steady state performance, high precision and fast tracking ability.