Reaerch On Permanent Magnet Synchronous Motor Servo Driver Based On DSP

In the modern automatic production control, the advantages of the permanentmagnet synchronous motor (PMSM), such as small volume, light weight, highefficiency, and a fall in the permanent magnet material price, makes the application ofPMSM is more and more widely. PMSM has gradually replaced the dominantposition of DC motor in the servo control system. Therefore, the control performanceindex of the PMSM has become the hot topics in the study of the current servo controlsystem. The thorough research of PMSM servo drive has been carried on in thispaper.The basic principle of PMSM is detailed in this paper. Meanwhile, themathematical model of PMSM which is deduced and discussed in depth is establishedunder three different kinds of coordinate system. The control system of PMSMemploy the control strategy with the magnetizing current is zero (id=0). The controlprinciple of voltage space vector pulse width modulation (SVPWM) and itsimplementation method is analyzed deeply. The simulation model of servo controlsystem is established with MATLAB/simulink. The simulation study includes thefollowing three aspects. Firstly, the state-space equations of PMSM are establishedthrough the mathematical model of PMSM. The double closed-loop control of servosystem is accomplished using traditional PI regulating method. The transfer functionmodels of electric current loop and speed loop of servo control system are designedand implemented. The system is approximated to a lower order system through thesimplified transfer function model. Then, the parameter of PI regulator is calculatedby the design methods of PI regulator. Secondly, the nonlinear mathematical model ofPMSM in two-phase stationary coordinate system is linearized. The simulationresearch of PMSM sensorless control is accomplished using the Extended KalmanFilter (EKF) algorithm which is used to estimate the rotation angle and therotation speed of the PMSM rotor. Thirdly, the adaptive control system of PMSMis established by combining neural network and PI control. The real-time training andlearning to neural network with feed forward is used of EKF algorithm which can overcome the defects of traditional BP algorithm with long training time and localminima. The adaptive control system of servo system is implemented by the PIDparameters adjusted with feedforward neural network on-line training.Finally, the design of hardware and software with servo control system iscompleted by using TMS32OF2812DSP chip made in TI Company as main controllerand high-speed chip of CPLD as auxiliary controller based on the requirements ofdesign performance.