Research On Fixed Point Control Of A Multi-vectored Propeller Aerostat Based On LPV Model

The multi-vectored propeller aerostat proposed in this paper was characterized by its low speed,nonlinearity,strong coupling and time-varying characteristics,making its pointing control different from others.The LPV model of the aerostat was considered,using scheduling parameters to reflect the system time-varying and non-linearity.The fixed-point and anti-windup controller design for the aerostat were addressed,adopting robust gain-scheduling and anti-windup control methods.The paper was organized as follows:Firstly,due to the strong degeneration and nonlinearity at low speed,the LPV was more accurate than the small perturbation model.The force analysis of the aerostat was carried out,and a nonlinear model was established.Then,the small perturbation equations of the aerostat was acquired at the equilibrium point,where the nonlinear model was decoupled in longitudinal and lateral directions.By fitting the linearized equations,the LPV model of the aerostat depending on the scheduling parameter was obtained.Secondly,the scheduling parameter was meshed in the work area as [2,6]m/s,the obtained values were stored in the form of tensor and a method of high-order singular value decomposition was applied.Thus the LPV model was transformed into the form of convex,with the vertexes and the corresponding weight functions given by MATLAB.A robust gain-scheduling controller was then proposed to maintain the aerostat in a fixed-point and simulation in MATLAB was carried out to verify the method.Finally,in view of the weak control ability of the stratospheric aerostat and the limited propeller thrust at low speed,a description of the control problem of the saturated LPV system was given.A method of parameter-dependent Lyapunov function was used to conduct the stable analysis of the system and the optimal estimation of the stable region was given.In the end,a synthesis of the saturation compensation controller and the optimal estimation of the domain of attraction was conducted.Simulation in MATLAB was carried out to verify the method.