Research On The Theory Of The 3-DOF Helicopter Model

Do low-altitude, low-speed and hovering maneuvers are the prominent feature of the helicopter, especially in a small area it can takeoff and landing vertically. These features make the helicopters have broad uses.It has been widely applied in military or civilian pespects. Helicopter flight control system is a typical multiple input - multiple output system, and has a strong channel coupling characteristics and nonlinear properties, it is a complex objects in the field of control engineering. Laboratory helicopter then developped for it can simulate part of the real helicopter's attitude and flight, so it is widely used in control theory of teaching, scientific research and so on.This dissertation did the research on control arithmetic and theory based on the 3-dof helicopter system produced by the googol tech Ltd.First, analyzed the helicopter system's composition .Modeling according to the characteristics of each degrees of freedom motion and selecting appropriate state variables to get the system state equation .Then designed PID controller to the each degree of the system; designed decoupling controller to the three degrees; Designed LQR controller based on the state equation we got before.Then genetic algorithm (GA) and particle swarm optimization (PSO) were introduced. GA was been used to optimize the PID parameters and LQR parameters. PSO was been used to optimize the PID parameters and compared the result with that was optimized by GA.Traditional control theory is based on the accurate model of the controlled objects. When it comes to complex system such as the helicopter system, the traditional theory is not applicable. So we introduced neural network to deal with such problem. In this paper, we used the RBF neural network to design controller. We used RBF neural network as identifier and controller, using RBF neural network to optimize PID parameter and one RBF used as identifier, both structures were built. We proposed that using PSO to optimize the initiate parameters of the neural network for there were no law to refer. The two types of controller simulation were made, and compared the effects of these two kinds of the control system.The feasibility of a theory must be tested for certification through the physical. Finally, we introduced the real time theory and carried out experience in Matlab under the control simulation. S function written based on the previously mentioned RBF optimize PID parameter control structure. In the Simulink environment, real-time control simulation was experienced. Experiment results show that the algorithm is feasible, and can get an ideal control.