Multivariable Identification And Control Of Multi-Rotor Based On Asymptotic Method

Quadrotor is one of the multi-rotor UAVs.It has the advantages of high mobility and low cost.It has been widely used in military reconnaissance,environmental monitoring,traffic supervision,disaster search and rescue and other fields in recent years.With the research of complex problems such as formation cooperative flight,higher requirements are put forward for the control performance of quadrotor.Because of the characteristics of coupling,under-actuated and nonlinearity,it is difficult to obtain accurate principle model.In order to improve the control quality,this paper will introduce system identification method for quadrotor,and design H∞ controller to improve the control performance.The main contents and contributions of this paper are summarized as follows(1)The asymptotic identification method is used to improve the previous results of quadrotor identification and control.A closed-loop identification test is designed,ensuring the stability of the system.The quadrotor attitude model is estimated and analyzed.And the test signal is under the persistent excitation condition.Considering the errors and uncertainties of the estimated model,a multi-variable H∞ controller for attitude angular velocity is designed.And the control problem is solved based on linear matrix inequality(LMI)method.Compared with the original controller,the attitude angular velocity tracking error is reduced by about 40%(2)The quadrotor system is a cascade control structure.There are four loops for attitude and position control.The identification experiment is modified to make the test data of the main and secondary loops be under the persistent excitation condition.And the controller for cascade structure is designed.Then the attitude angle model and position model of quadrotor are estimated and controlled.The flight experiment results show that the control quality of attitude angle and position is improved.(3)Considering the nonlinearity of the quadrotor attitude model,the linear parameter variable(LPV)model is used to estimate the angular velocity model of the quadrotor.Taking attitude angle as scheduling variable,the linear models with different angle ranges are identified as sub-models,and then LPV model is obtained by iteratively optimizing the weighting function and sub-model parameters with the whole data Then,based on the vertex property of LPV model,the H∞ gain scheduling controller of the quadrotor attitude model is designed.From the simulation results,the control performance is improved compared with the linear H∞ controller designed before.