程序调参飞行控制律的神经网络实现

为保证飞行品质，现代飞机中多采用程度调参飞行控制律，由于调参规律的高度非线性，使得控制律的工程实现相当困难。为此，对程序调参飞行控制律的神经网络实现方法进行了研究。通过应用ＢＰ神经网络逼近非线性调参规律。探讨了用ＢＰ网络实现程序调参飞行控制律的一般方法，给出了网络的结构及其优化算法，并以某型飞机电传操纵系统倾斜阻尼通道的程序调参控制律为例，对该实现方法进行降验证。结果表明，实现精度符合工程要求。

Neural Network Realization of Gain Scheduling Flight Control Law

In this paper, the neural network realization method of gain scheduling flight control law is mainly studied. In order to obtain high flight qualities, gain scheduling control laws are widely used to adapt variations of dynamic models of airplanes in modern airplanes. Due to the high non linearity of gain scheduling functions, it is difficult to realize the control laws in engineering. BP neural network has outstanding abilities of processing nonlinear function. By using BP network to approach the gain scheduling function, the high non linearity is overcome in this paper. The network structure and optimal algorithm are given. By using 3 layer BP network, this kind of realization method is verified through its application in the damped lateral channel of an aircraft's fly by wire system, the average square error of network structure is smaller than 0.6%. The result shows that the realization precision is high enough for engineering application.