Research On Drive Technologies Of Dual Three-Phase Permanent Magnet Synchronous Motor

Compared with conventional three-phase motor, multiphase motor has the advantages of large power output with low voltage, good fault tolerant ability. It is suited for some application areas, such as electric ship propulsion, electric vehicles and aircrafts. With the development of power electronics and micro-electronics, phase number is not a restriction in the system design. Multiphase drive system has attracted more and more attention. Dual three-phase permanent magnet synchronous motor (PMSM) which has two sets of three-phase windings phase shifted by 30 electrical degrees is chosen as research object. Its drive and control technique are deeply studied in this dissertation.Firstly, the models of dual three-phase PMSM are set up in the double d-q and vector space decomposition (VSD) reference frames. Two vector control strategies related to the two models are proposed. The double d-q vector control is equivalent to the control of two three-phase motors. In VSD model, the current is controlled in two orthogonal subspaces. The comparative analysis of the two strategies shows that they have consistent control performance of torque and speed when their control parameters have the same value. The VSD model indicates the features of multiple harmonic subspaces of dual three-phase motor, so its current control is more flexible. With the same speed control performance, VSD vector control strategy can use four-dimension current control to minimize stator loss or two-dimension current control to simplify the system control structure.Secondly, the pulse width modulation (PWM) algorithms for dual three-phase motor are analyzed. Conventional two-vector space vector PWM (SVPWM) has a high DC bus utilization, but the output voltage has large harmonics. The four-vector SVPWM is just used to produce sinusoidal phase voltage, so the DC bus utilization cannot reach to the highest value. Four-vector SVPWM is extended to the non-sinusoidal voltage range in this dissertation. It has the same modulation range as the two-vector SVPWM, but reduces the voltage harmonics in non-sinusoidal voltage range. So the current harmonics and stator loss can also be reduced. The four-vector non-sinusoidal voltage modulation algorithm is then integrated into double zero-sequence injection PWM and three-segment modulation is proposed. The total modulation range is divided into three segment, sinusoidal current range, sinusoidal voltage range and non-sinusoidal voltage range. The first is related to four-dimension current control and the last two are related to two-dimension current control. This modulation algorithm has an optimal utilization of DC bus in the whole modulation range.Thirdly, the nonlinear control strategy for PMSM is studied. A fully decoupled linear model of PMSM is obtained by output feedback linearization. In this model, the speed can be controlled by directly controlling the voltage. The performance of the feedback linearization control depends on the parameters very much. To solve this problem, terminal sliding mode (TSM) control is used to design the speed and direct current controllers. The control laws are deduced and the system stability is proved. In order to reduce the sliding mode gain and chattering phenomena, a disturbance observer is designed. The disturbance observed is then compensated in the control. The simulation results show that the feedback linearization terminal sliding mode control with disturbance compensation improves the dynamic performance of speed and has good robustness.Lastly, in connction with the feature of multiple control dimensions of dual three-phase PMSM, fault tolerant control strategies with open phases are researched. Different neutral connections correspond with different optimal current control methods when the multiphase motor has open phases. Vector space decomposition transformation matrixes are determined according to different neutral connections and their current constrains with one open phase or two orthogonal open phases. Then different VSD models are deduced. All the variables are projected into d-q subspace which has relation to the electromechnical energy convertion and the z1-z2-z3 subspace which has no contribution to electromechnical energy convertion. Vector control methods based on VSD models are proposed. By setting different current references in z1-z2-z3 subspace, two optimal current control methods of minimum stator loss control and minimum current magnitude control are obtained. The effectiveness of the fault tolerant control methods are confirmed by simulation and expermnet results.