Intelligent Control On Ducted Fan Unmanned Aerial Vehicle Based On Attitude Guidance And System Identification

Ducted fan unmanned aerial vehicle (DF-UAV) is widely used in high-altitudesurveillance for its excellent maneuverability and high propulsion efficiency.However, DF-UAV is easily disturbed by wind gusts which is uncertain andtime-varying. And this is the reason why a robust controller is needed. This paperbuilt the framework of flying controller for DF-UAV and studied how to identify thedisturbance to reach the goal of hovering and path following and solve the problemsmentioned above.At first, system modelling. Given the dynamic property of propeller and controlsurfaces, treat DF-UAV as an11degree of freedom (DOF) under-actuated nonlinearsystem. Cardan angle and transition matrix was used to describe the attitude inmotion analysis. From the Lagrange-D’Alembert principle, the dynamics ofDF-UAV was then derived. As to the forces and torques acting on the DF-UAV,introduce aerodynamic lift and drag caused by the motion of propeller and controlsurfaces.Secondly, layered controller design based on virtual force guidance. Accordingto the requirement of control system, the top layer was to design the pseudo controlinputs. Then reference feedback commands that incorporate the compressedinformation of pseudo control inputs was induced from dynamics of DF-UAV.Meanwhile, Control information was passed down to the propeller and controlsurfaces commands through dynamic decoupling algorithm. And it was the controllaw for the propeller and control surfaces at the bottom layer.Thirdly, attitude-guided path following control design. Through the analysis oftranslational dynamics, attitude motion was assumed as a virtual input of the lateraland longitudinal flight dynamics. Then design of attitude command generatorseparated attitude and flight dynamics. Given that the dynamics of DF-UAV wasuncertain and DF-UAV was easily disturbed by wind gusts, a method of systemfunction approximation based on sigmoid function and state variable integral wasproposed in the process of controller design. Stability analysis showed uniformultimate boundedness of errors of path following and approximation. Comparison ofthe control laws mentioned above was made to show the superiority of intelligentapproximation algorithm.At last, co-simulation between MATLAB and Adams was conducted to verifythe feasibility of control system mentioned above. The simulation results was inaccord with the theoretical analysis. As to the ability of resisting wind gusts, the firstcontroller was not that good when compared with the second one, which illustrated the importance of the system identification method.