基于小波神经网络的飞行控制律设计

Flight Control Law Design Method Based on Wavelet Neural Network

研究战机稳定性能优化挖掘问题,由于飞行控制器增益要求实时调整,达到飞行稳定。为解决上述问题,提出了一种小波神经网络的大包线增益调参飞行控制器设计,根据神经网络的非线性泛化能力获得各个学习样本飞行状态间控制器参数的取值,利用小波神经网络来实现大包线增益调参飞行控制,可以提高收敛速度与逼近精度,并可利用免疫粒子群算法替代BP算法对小波神经网络中的参数进行优化。仿真结果表明,所设计的小波神经网络增益调参飞行控制器实现了大包线飞行控制,并且具有良好的稳定控制性能。

A large envelope gain schedule flight control law design method based on wavelet neural network （WNN）was propsed in this paper. Because of generalization of WNN, it can well calculate the gain between the training points and realize the flight control. In order to improve the calculating speed and the nonlinear function ap-proximation capability, an immune PSO algorithm was used to train the parameters of the wavelet neural network. The simulation results show that the WNN gain schedule flight controller can relaize the large envelope flight control and has good performance.