Three Degrees Of Freedom Aircraft Model-based Control Method

As a traditional controller, PID controller is characteristic of simplestructure, and have good adaptability and extensive applications. However,it can't get satisfied, especially for complicated systems which are ofcharacteristics of nonlinearity and time-varying in the industrial controlfields. Involving neural network and PID controller in an organic whole,the NN—PID controller has a good control effect, which has the merit of anyPID controller for its simple construction and definite physical meaning ofparameters, and of the self-learning and adaptive functions of a neuralnetwork.The 3D hover model system, provided by the Canadian company of Quanser,is simple and intuitionistic as a laboratory setup but is complicated as acontrolled unit. It is a high order, unstable, multi-variable, non-linear andcross-coupling 3-DOF multi-input multi-output complex system, which can bestable by adopting effective control method. So, in the paper, aself-adjustment PID controller based on RBF neural network was studied indetail. The work is listed as follows:Firstly, models are developed in this thesis. Multi-Input andMulti-System Models of the Helicopter are developed in Matlab, including modelof the horizontal part of system, model of the vertical part of system, modelof the pitch part of system and 3-DOF model of system.Secondly, all kinds of usual state space design methods of traditionalcontrol theory are introduced, focusing on PID control and optimizationcontrol system with LQR controller. The methods are simulated in Matlab.Finally, based on radial basis function neural network, a PID controllerbased on RBF network is designed for 3D model hover. The simulation resultindicates that the system, compared to the LQR control method, possesses theadvantages of high precision, quick response speed and is of greatadaptability and robustness.