Research On Control Of Low Wind Speed Magnetically Suspended Vertical Axis Wind Turbin

Low-wind-speed wind power is one of the key developing areas in China’s future wind power.However,in low-wind-speed areas,wind energy density is low,wind direction changes frequently,and turbulence is strong.Existing horizontal axis wind turbines have inherent defects such as yaw device needed and high starting resistance moment,which makes it difficult to meet the requirements of low-wind-speed wind farms.For this reason,our research group proposed a new low-wind-speed maglev vertical axis wind turbine(MVAWT),which mainly consists of maglev disc motor(MDM),main generator,i.e.permanent magnet direct-driven wind generator(PMSG)and converter system.This paper focuses on the control of MDM and its maglev converter(MC),as well as the control of PMSG and its machine-side converter(MSC).The main contents are as follows:First,for the MDM and its MC,a finite-time suspension control strategy based on global fast terminal sliding mode(GFTSM)and adaptive RBF neural network(RBFNN)is proposed.Based on the maglev dynamic mathematical model with uncertain disturbance(such as random changes of wind speed and wind direction),a finite-time GFTSM controller with small chattering is designed.An adaptive RBFNN is used to approximate uncertain unknowns of the system,and the finite-time fast convergence characteristic of GFTSM is introduced into the adaptive law to improve the convergence speed of RBFNN.Furthermore,in order to obtain the derivatives of input signals of controller and mitigate the effect of sensors’ noise,a finite-time differentiator is introduced.Finally,the stability and finite-time convergence characteristic of the proposed control strategy are proved according to Lyapunov stability theory,and the simulation and experimental platforms are built respectively for research.The results show that the maglev system controlled by the proposed strategy has faster dynamic response speed and stronger anti-disturbance ability.Second,for the PMSG and its MSC,a deep reinforcement learning control strategy based on Actor-Critic architecture and RBFNN(AC-RBF)is proposed.In order to solve the problems that the parameters of traditional PID controller are not easy to adjust,and difficult to cope with the complex time-varying environment,a PID speed tracking controller based on AC-RBF(AC-RBF-PID)is proposed for PMSG.The 3-5-4 strcture of RBFNN is used to adjust the incremental PID controller parameters.The RBFNN is adopted to output the control parameters to be adjusted and approximate the value function simultaneously.In addition,the zero d axis current decoupling control strategy is employed to solve the coupling problem of stator current.Furthermore,the feasibility of adjusting PID parameters by deep reinforcement learning AC-RBF is verified by simulation,as well as its superiority for the control of PMSG.Finally,considering that the uncertainty of MVAWT external disturbance is prominent in low-wind-speed areas,the parameters of the PMSG and its converter will also deteriorate due to the devices aging or long-term operation,and the mentioned above decoupling error is inevitable,so in order to further improve system performance,another control strategy based on deep reinforcement learning TD3 is proposed for PMSG and its MSC,and the optimal controller is obtained by designing and training TD3 agent.And the research is carried out according to the simulation platform and the experimental platform.The results show that the dynamic response of PMSG controlled by the proposed TD3 control strategy is faster and the overshoot is smaller than that controlled by AC-RBF-PID and traditional PID.Moreover,the proposed TD3 control strategy can suppress the current harmonics and reduce the electromagnetic torque ripple,so as to effectively reduce the power loss of MSC of PMSG and improve the power generation efficiency.