Research On Sensorless Control Technology For Permanent Magnet Synchronous Motor Speed Control System

Permanent magnet synchronous motor(PMSM)has been widely used in the automatic control of industrial areas,transportation,aerospace engineering,etc.,due to the advantages of simple structure,excellent dynamic performance,energy saving,and so on.High-performance PMSM speed control system needs rotor position and speed information from the mechanical sensor installed on the axis of rotor to develop a closed-loop speed control system.However,the mechanical sensor not only increases the weight and the cost of the system,but also decreases the system reliability and restricts the applications in some special areas.To remove the mechanical sensor and adopt the sensorless control will bring important practical value and economic benefits.Nonetheless,the nonlinearities of pulse-width modulation voltage source inverter(PWM-VSI),the inaccurate motor parameters and the nonideal motor etc.will lower the accuracy of estimated position and degrade the performance of speed control system.To solve these problems,these factors are analyzed and some strategies improving the dynamic and static performance are proposed.The main researches and innovations of the thesis are on the following areas:PWM-VSI,such as dead-time,voltage drops across switches and diodes,turnon/off time delays of switches,etc.,lead to output voltage decrease,voltage and current distortion,torque and speed ripple in drives.The models of the above nonlinearities are established and the influence can be treated as disturbance voltages in the synchronous rotating frame.The influence of these disturbance voltages on the current loop and speed loop is analyzed,and a novel repetitive learning controller(RC)in an angle domain is proposed to suppress this disturbance.Meanwhile,the choices of the parameters and stability of the RC are analyzed.The simulation and experimental results demonstrate this strategy is less dependent of the system model,easy to adjust the parameters,and improve performance of PMSM speed control system effectively.The sensorless control of PMSM speed control system need accurate motor parameters.An unscented kalman filter(UKF)for multiple parameters estimation of PMSM is proposed to estimate the stator resistance,flux linkage of permanent magnet,d-axis inductance(d-axis)and quadrature-axis(q-axis)inductance simultaneously.These estimated parameters are regarded as states of nonlinear system.Hence,the parameter estimation is transformed to the state observation.The sufficient condition of the observation of four parameters is analyzed.Through the injection of the exciting current in the d-axis current reference,the observation matrix of nonlinear system is full rank.Meanwhile,through the compensation of nonlinearities of PWM-VSI,the accuracy of estimated parameters is improved.A comparison of UKF,extended kalman filter(EKF)and recursive least square algorithm is carried out in terms of convergence domain,accuracy,speed of convergence and anti-noise both in simulation and experiment,and the pros and cons are summarized.The simulation and experimental results show that the UKF can give accurate parameters estimation and these parameters can be used in sensorless control of PMSM.To develop a simple and reliable whole speed range sensorless control of PMSM,a whole speed sensorless control of PMSM considering the magnetic saturation effect is proposed.PMSM starts up using the I/F method,then switches to the UKF sensorless control.A switching method combined with a current switch and a position switch is implemented to realize current and speed changing smoothly.Considering magnetic saturation effect on sensorless control in middle and high speed range using UKF,the varying inductance models are developed,meanwhile,filter delay compensation is proposed.The experimental results demonstrate that accuracy of estimated position and dynamic performance are improved.What's more,a comparison between UKF and EKF sensorless control is carried out.The errors of estimated position in the high frequency injection method—rotating high frequency voltage injection(rotating method)and pulsating high frequency voltage injection(pulsating method)are analyzed,which can be grouped into two categories: the errors due to the extraction of the position information from the high frequency current and the errors due to the nonideal of the PMSM.The influence of stator resistance,filters,discretization,phase-locked loop on the errors of estimated position using rotating method and pulsating method is analyzed.What's more,the compensation of these errors and an UKF direct position estimation method are proposed.Meanwhile,the mechanism of magnetic cross-saturation effect and multiple saliencies effect and the influence on the errors of estimated position are analyzed,and the compensation of magnetic cross-saturation and the decoupling method based on RC in the angle domain are proposed.The simulation and experimental results show the proposed method can improve the accuracy of estimated position and estimated speed,and improve the performance of PMSM speed control system.