超机动飞行的神经网络动态逆控制

根据反馈线性化理论，讨论了神经网络自适应非线性动态逆控制设计。首先根据时标分离的原则，采用动态逆方法设计了快回路和慢回路控制器；其次提出了模型的神经网络非线性直接自适应控制方案，其中设计一种在线神经网络用于补偿模型逆误差。仿真表明，该控制方案具有较好的自适应能力的鲁棒性。

Neural Network Dynamic Inversion Control and Simulation of Supermaneuverable Flight

A discussion is devoted to the design of a self adaptive and nonlinear dynamic neural network inversion controller according to the feedback linearization theory. The control of uncertain nonlinear systems is an active subject in the modern control area, and there are a lot of research results about them. Neural networks(NN) are widely used to control nonlinear systems, so a lot of control methods have been proposed. But most of them are control methods of single input and single output nonlinear systems. There are few control methods of multi input and multi output uncertain nonlinear systems. The discussion in this paper is focused on multi input and multi output uncertain nonlinear systems. Firstly, with two time scale, the design of a fast loop controller and a slow loop one is given by using the dynamic inversion method. Secondly, a neural network nonlinear, direct and self adaptive scheme based is presented a model inversion, in which an online neural network is designed to compensate the model inversion errors. It is demonstrated by the distributed simulation of a local area network that this adaptive control design has good adaptability and robustness.