Research On Multivariable Control Algorithm Of Electric Propulsion Power System Of UAV

Electric propulsion power systems have been widely used in small and medium-sized UAVs,among which the variable-pitch electric propulsion power system is a research hotspot in this field.In this thesis,a typical variable pitch electric propulsion power system is token as research object,and a testbench for the variable-pitch electric propulsion power system is built.Then,the nonlinear model and LPV model of the variable pitch electric propulsion power system are established by component-level modeling and parameters identification.The model correction algorithm of variable-pitch electric propulsion power system,nonlinear decoupling control and linear control algorithm of multi-variable power system are mainly explored.,the specific research contents are as follows:(1)In order to solve the problems of performance degradation of the electric propulsion system due to battery charge attenuation,and of the state of airborne LPV model and output deviation,the degraded variable pitch electric propulsion system was simulated using nonlinear model and the discharge SOC curve of lithium battery.Then the airborne LPV model is corrected by linear Kalman filter and extended state observer.The results show that the tuned airborne LPV model can track the speed and lift of the electric propulsion system very well.(2)Aiming at the non-linear and multi-variable coupling characteristics of the power system,the nonlinear model was taken as the algorithm verification object,the online feedforward decoupling control algorithm based on the onboard LPV model and the BP neural network multi-variable adaptive decoupling control algorithm are proposed.The results show that,compared with the traditional feedforward decoupling control,the online feedforward decoupling control based on LPV model initially solves the problem of decoupler adaption,while the BP neural network multivariable adaptive decoupling controller can simultaneously self-tune the PID parameters online of the multiple control loops according to the change of operating conditions,meanwhile ensuring the adaptive capability of decoupler and controller.(3)Aiming at the disturbance rejection problem of the power system near the fixed operating point,the linear model of the power system is taken as the algorithm verification object.To suppress the disturbances suffered by the system,the LQR controller is used which parameters are optimized by PSO algorithm.In addition,due to the LQR controller is more dependent on linear model,the problem of linear model parameter perturbation cannot be ignored.To solve the forementioned problem,the model matching condition and the Lyapunov function were utilized to design the control law of the MRAC algorithm.The results show that,compared with the traditional LQR control,the LQR controller optimized by PSO algorithm has a stronger disturbance rejection capability.And after using the MRAC algorithm,the state of the power system can always maintain an asymptotic tracking to the state of the reference model,which solves the problem of parameter perturbation of the linear model and further strengthens the robustness of the linear control system.