The Study on Attack Helicopter Adaptive Flight Control System Using Neural Network
|School||Nanjing University of Aeronautics and Astronautics|
|Course||Pattern Recognition and Intelligent Systems|
|Keywords||Neural Networks Adaptive control Helicopter Flight Control System Nonlinear systems Dynamic inversion Wavelet Neural Network Maneuvers|
In-depth study and exploration of adaptive neural network control in helicopter flight control system, the main contents include helicopter flight control system design of dynamic inversion, wavelet neural network dynamic inversion error compensation, helicopter maneuvers simulation, and other aspects of the content. Helicopter there is a strong coupling, instability, showing a strong nonlinear Firstly helicopter nonlinear equations of motion, and free open-loop characteristics of the aircraft research and analysis of the movement of the coupling and manipulation of the impact of the coupling on the respective states of the helicopter, and to expose the coupling between the various channels of the helicopter, as in the following control system design to provide certain physical concept. Channel helicopter attitude angles and pitch channels were designed. The attitude angle channel design, the use of dynamic inverse thinking, helicopter flight control system in accordance with the principle of separation of the time scale is divided into fast and slow circuit, and were used dynamic inversion method for design, simulation, and three attitude angle channel helicopter been decoupled. The inverse dynamic inverse controller error caused by imprecise model, using wavelet neural network compensation. This paper discussed the wavelet theory, given the general structure of the wavelet function. The inverse dynamic inverse error first BP wavelet-based neural network research, learning algorithm and the right value adjustment law designed neural network controller; then based on Lyapunov stability wavelet neural network research on the neural network adaptive control law detailed analysis and discussion of the neural network controller is designed. The neural network controller design of these two methods of simulation show that they can be good compensation dynamic inversion generated error. Finally, the design of the helicopter fuzzy the PD trajectory tracking controller, combined with the attitude angle controller to complete the design of the flight control system. And two typical maneuvers (in the vertical plane, anxious jumped / Jiyue the drop, the S-shaped maneuver in the horizontal plane) for digital simulation results show that the control system designed in this article to improve the mobility of the helicopter.