Forward Neural Network In The Vehicle Longitudinal Control System Applied Research,

Intelligent Control based on neural networks is the powerful method for the nonlinear systems. In this paper multilayer feedforward networks is used to design the nonlinear flight control systems, including the systems identification and control. The main contribution and creativeness result of this dissertation are summarized as follows:Firstly, the basic theory of feedforward neural networks is reviewed. And this paper summarized the identification and control methods of nonlinear based on neural networks. Using the approximate ability to nonlinear function of neural networks, the adaptive frame of identification and controller are constructed. And the formula of Jacobin matrix for unknown systems is given, which is used to train the neural networks controller in closed systems.Secondly, the adaptive identification frame is constructed and applied to the missile systems which contains various nonlinear, unknown factors. And the main parameter model is also constructed using neural networks. Neural networks is used to identify the aerodynamic characteristic of the missile. The effectiveness of neural identification is verified by simulation study.Finally, the adaptive inverse controller based on neural networks is applied to keep longitudinal incidence. And the adaptive parameter controller is applied to missile autopilot, which controls the overload of missile. The effectiveness of neural controller is verified by simulation study.The study in this dissertation is of theoretical meaning and engineering application significance for the unknown nonlinear systems.