INVESTIGATIONS ON PERMANENT MAGNET SYNCHRONOUS MOTOR DRIVE

Abstract

Adjustable speed drives are essential for energy savings in industries. Traditionally DC motors have been used for the adjustable speed drives. Now-a-days vector controlled ac drives are replacing the dc drives because ac motors are more robust and need minimum maintenance due to absence of commutators and brushes. Among the ac motors permanent magnet synchronous motors (PMSM) are superior to squirrel cage induction motors for servo applications as it has fast response, higher torque to inertia ratio, higher power density and higher efficiency. Further, the permanent magnet ac motors do not need magnetizing current from stator side. The magnets on rotor are of high resistivity materials thus losses in the rotor are negligible. Due to absence of the squirrel cage, the rotor is lightweight and small in volume. The stator of the PMSM is provided with a sinusoidally distributed three-phase winding and the rotor is fitted with permanent magnet which is chamfered so as to produce sinusoidally distributed magnetic flux in the air gap. A good amount of work is reported in area of control of the PMSM drive. In most of control techniques, design of controllers depends on parameters of the motor and load. As the parameters vary with saturation, temperature and operating conditions, it is difficult to get optimal design of the controllers. The drive performance also deteriorates with the motor and load parameter variations. Thus, it is required to have robust controllers, which are independent of the motor and load parameters. Further, in a conventional PMSM drive an uncontrolled three-phase rectifier is used as a front-end converter. The rectifier draws non-sinusoidal current from the three-phase utility thus injecting large amount of current harmonics. During retardation of the drive, kinetic energy of the rotor is converted into electrical energy and it increases dc link capacitor voltage of the rectifier. To avoid this a regenerative front-end PWM converter is required. The drive scheme under investigation consists of a front-end regenerative PWM converter and a PWM inverter on the PMSM side. The drive power is drawn from the three-phase utility by the PWMconverter and a regulated output voltage is produced at its dc link capacitor. The converter draws sinusoidal input current at unity power factor, thus injecting negligible current harmonics into the utility. The PWM inverter draws power from the dc link terminal to produce variable frequency and regulated three phase ac currents for the PMSM. The drive has two independent control structures; one is inner loop current control and outer loop voltage control for the front-end PWM converter and the other is inner loop current control and outer loop speed control of the PWM inverter fed PMSM. Hysteresis current controllers are chosen for the inner loop current controls of both the PWM converter and the PWM inverter fed PMSM. The dc link voltage of the PWM converter is sensed and compared with a reference dc link voltage; resulting voltage error is processed in a PI voltage controller, which outputs peak amplitude of reference current. The peak amplitude reference current is multiplied with three-phase sinusoidal unit vectors to obtain the three-phase input reference currents. The three-phase sinusoidal unit vector templates are extracted from the utility phase voltages by dividing with their peak amplitudes. The hysteresis current controllers make actual three-phase input currents to track the respective phase reference currents, thus ensuring unity power factor operation of the converter. In the PWM inverter fed PMSM control structure, the rotor position and speed are sensed. The actual rotor speed is subtracted from a reference speed and speed error is processed in speed controller to obtain a reference torque. The torque command is divided with the PMSM torque constant to obtain torque component (q-axis component) of the reference current. The field component (d-axis component) of the reference current is kept zero as the PMSM is surface mounted type in which demagnetisation is avoided. Using torque and field components of the reference current and rotor position angle, the Park's transformation is applied to transform dqO-axis reference currents to abc-axis reference currents. The actual three phase stator currents are subtracted from the abc-axis reference currents and current errors are passed through hysteresis current controller, which provide controlled switching signals to the inverter switches. The controlled switchings of the inverter feeds regulated current to the PMSM. With speed and current control loops, four-quadrant speed control of the PMSM is achieved. With retardation command of the motor speed, kinetic energy of the rotor is converted into electrical energy and it is fed back to the utility. At retardation the dc link current reverses and the PWM converter goes from rectification mode to regeneration mode. Mathematical models of the PMSM, PWM inverter and PWM converter are developed. The models are used for the associated controller designs and simulation of the drive. The inner current control loops of the converter and the inverter are executed with hysteresis current controllers. For the design of the hysteresis current controller, the system parameters are not required it needs only bandwidth information. The PMSM speed controller is investigated with PI speed controller and fuzzy-PI speed controller. The drive performance is compared with the PI and fuzzy-PI controller. Design of the fuzzy-PI controller does not need any system model parameters. The design of PI speed controller uses load-torque parameters. To design the speed controller, the current controller of the inner loop is assumed ideal. The speed controller is processed on a digital processor, hence, the speed loop control structure is analysed in discrete time domain. The closed loop transfer function of the drive is obtained in discrete time domain. Stability of the closed loop PI speed control system is studied with variations in sampling period, damping friction and load inertia. Choosing a relatively higher stable region on the controller gains plane, optimum value of the controller gains is obtained using Genetic Algorithms search method. The outer loop control of the PWM converter is PI voltage controller. The inner current loop is assumed fast enough to delink it with the outer loop, while designing the PI voltage controller. Power input and output balance dynamics is used to obtain current to voltage transfer function of the converter. The voltage PI controller is processed on a digital processor, hence; closed loop transfer function of the outer loop of the PWM converter is also obtained in discrete time domain. With the closed loop transfer function of the converter, its stability is observed in the proportional and integral gains plane. Again, Genetic Algorithms search method is used to obtain optimum PI controller gains from a relatively higher stable region for a desired voltage response. Once, the voltage and speed controllers are designed and bands are chosen for the hysteresis current controllers, the complete drive can be simulated. Transients as well as steady state performances of the PWM converter and the PMSM drive are observed with the optimised voltage and speed controller gains. The performance of the hysteresis current controllers play important role on producing sinusoidal input currents of the converter and the PMSM. A good current controller produces minimum current harmonics with less number of switchings per cycle. A modified hysteresis current control method is proposed. Both the PWM converter and the PWM inverter fed PMSM current controllers are implemented with conventional as well as with the modified hysteresis current controllers. Performances of both the current controllers are observed under same operating conditions. With the modified hysteresis current controller fewer number of average switchings per cycle, more sinusoidal input current and less total harmonic distortion in the currents are observed. Due to more sinusoidal input currents in the PMSM with the modified hysteresis current controller, less torque ripples and less steady state speed errors are observed. A prototype model of the PMSM drive along with the front-end three-phase PWM converter is developed. The current controllers are implemented using analog circuitry. The voltage controller of the PWM converter and speed controller of the PMSM are processed on a digital computer with floating point calculations. The rotor positions are sampled using a resolver through analog to digital converters (ADCs) and the reference currents are obtained through digital to analog converters (DACs). The reference speed and the reference dc link voltage are entered through keyboard of the computer. The controllers are processed in a hardware-interrupted subroutine. The off-line optimisation of the speed PI controller parameters is done with simulated performance of the drive. The drive model used for simulation does not include core saturation and non-linearties present in PMSM load and inverter switchings. Hence, on-line optimisation of the speed PI controller gains is also done using Genetic Algorithms. With both the on-line and off line optimised controller gains speed response ofthe PMSM are observed and compared. The experimental results include speed controller response, voltage controller response, the conventional and modified hysteresis current controller response with the PWM converter and PWM inverter to investigate performance of the drive. PI speed controlled PMSM drive is not robust as its response depends on the motor load parameters. Arobust speed controller is required to overcome dependency of the speed controller on the motor load parameters. A fuzzy-PI speed controller is designed and it is substituted in place of the speed PI controller. Output scale factor of the fuzzy-PI controller is kept equal to maximum value of the rated torque of the PMSM and input scale factors for speed error and change in speed error are optimised off-line using Genetic Algorithms. The simulation and experimental speed response of the PMSM drive with off-line optimised controller gains of the PI speed controller and with off-line optimised input scale factors of fuzzy-PI controller are compared. The performance of the speed PI controller depends on the PMSM load parameters. The inner current control loop is independent of the system parameters. The fuzzy-PI controller is also independent of the system parameters. Thus, with hysteresis current controller as inner loop control and fuzzy-PI controller as outer loop control, the PWM inverter fed PMSM drive becomes independent of system parameters. To summarize, a vector controlled current regulated PMSM drive with front-end PWM converter is designed and developed. PI controller gains for speed and voltage controf loop are optimised using Genetic Algorithms for desired performance index. A modified hysteresis current controller is proposed and its performance with the PWM converter and PWM inverter fed PMSM are investigated and compared with the conventional hysteresis current controller. Implementing the fuzzy-PI speed controller along with the hysteresis current controller, the PMSM drive control become independent of parameter variations.