The Study Of Quadrotor Aircraft Control System Based On PID Neural Network

Quadrotor aircraft is a new kind unmanned aerial vehicle which can verticallytake-off and land (VTOL). It has many advantages such as simple construction, largeloading capacity, strong flexibility and easy to achieve hover. So it is valuable for bothcivilian and military usages, especially suitable for applying to the task of low-levelinspections of complex terrain and the aerial photograph. Using self-made microQuadrotor as research platform, this thesis build the dynamical model and researchthe Quaternion attitude resolving method and the complementary filter design. Theco-simulation platform is built by using Adams and Matlab, and the traditional PIDcontroller and the PID neural network controller are set up innovatively. Therefore theobject of shortening development period, reducing the difficulty of setting PIDparameters and making the Quadrotor to hover and automatically cruise is achieved.Firstly, this thesis describes the research and development of quadrotor aerialvehicle at home and abroad in detail. Then according to system function and materialprice, choosing suitable engineering materials and MEMS electronic components toset up physical prototype and hardware control system of quadrotor. Based on thephysical platform, principle of quadrotor and dynamics features are analyzed, andNewton-Euler equations is used to build6DOF aircraft dynamics model whichprovides the basis for controller design.Secondly, compared to the common strapdown solution algorithm, quaternionalgorithm is selected to express the attitude calculation and establish the turningquaternion differential equation. Integrated with constructed inertial navigationattitude monitoring system and based on complementary filter algorithm, anoptimized filter is designed. The filter combined the data of gyroscope, accelerometerand electronic compass, current attitude of the quadrotor is accurately solved.Finally, the co-simulation platform is established by using the dynamics softwareAdams and the control simulation software Matlab/Simulink. Based on the platform,classical PID attitude controller is designed and step signal is input into for simulation.Through setting parameter repetitiously, the result shows that the designed controller is not good to the strong coupling nonlinear system. Then PID neural networktechnology is introduced in detail, and the PIDNN attitude controller and PIDNNposition controller are designed. The result of simulation shows that the algorithmachieve general decoupling control to the quadrotor aircraft system. The PIDNNalgorithm is proved to be effective and advantageous by the actual flight test andanalyzing the result.