Flight Control Of Quadrotor UAVs Based On Model Predictive Control

Quadrotor UAV has advantages such as simple structure,maneuverability,mobility,etc.With technical progress in recent years,quadrotor UAVs have been applied to various fields.In this paper,quadrotor's motion dynamics is analyzed to establish mathematical model.Two predictive controllers based on different models are designed and improved to realize flight control.Finally,a hardware-inthe-loop simulation system is built to verify control effect of the predictive controller.Firstly,a six-degrees-freedom dynamic model is established considering wind disturbance which contains motor dynamics,so it can be applied to actual platform and then get linearized.Then,a step response model is established based on the linearized model to design classical dynamic matrix control(DMC).Numerical simulation verifies its correctness and limitations when it is applied to flight control.The predictive control based on state space model is then studied.Simulation demonstrates its correctness and connections with linear quadratic regulator(LQR).Next,two parametric models are extracted to design flight controller.Predictive control algorithm get improved in three points: reducing of compulation amount,adding of physical constraints and optimized control on output deviation,the first point is achieved by characterizing the control sequence in laguerre function;the second point is achieved by solving quadratic programming in hildreth algorithm;the last point is achieved by introducing output deviation differential.Then facing the control needs in large-scale flight state,using different local linear model to construct linear expansion model and predictive model to obtain local wind disturbance observer and predictive controller,the disturbance observation and control effect far from hover state point get improved,then control output is compensated by interference value to get better quadrotor fight control effect.Last,a hardware-in-the-loop simulation system is built to verify control effect and engineering feasibility of the design.It is composed of simulator,ground station and flight controller.In order to have engineering guidance significance,this paper proposes an improved RRT algorithm as the core of the track generation module to verify the control performance in real application.The experimental results show predictive controller can achieve trajectory-keeping and attitude stabilizing control effectively,and it has anti-interference ability and high engineering feasibility.