Duty cycle model predictive torque control (Duty-MPTC) only uses zero vector as the second vector to form vector combinations, which restricts its torque control performance. This paper proposes a modified double vectors model predictive torque control (DVMPTC) by reasonably selecting active vector instead of zero vector. By analyzing the regulation of torque and flux of permanent magnet synchronous motor (PMSM), the torque and flux difference parameters are obtained, and then a second vector selection table is established to determine the best vector combination for torque ripple suppression. And based on the torque and flux difference parameters, the duty cycle calculation method is proposed to reduce the calculation burden of the control system. In order to verify the effectiveness of the proposed method, experiments with a 6-kW PMSM are carried out. Compared to Duty-MPTC, the proposed DVMPTC method can significantly reduce the torque ripples with good dynamic performance and relatively low average switching frequency.

In order to meet the high performance and efficiency requirements, a new space vector-based discontinuous modulation strategy for neutral-point clamped three-level inverter is presented in this paper. Different from the conventional discontinuous PWM (DPWM) strategies, a new clamping state is applied to synthesize the reference voltage under the condition of low modulation index. The rms value of output current ripple for different clamping states are calculated, and then the distribution of the new clamping state and the other two clamping states adopted in conventional DPWMs is optimized based on the calculation results. The switching sequences are redesigned according to the new distribution of the three clamping states. The output voltage and current, switching loss, neutral-point voltage ripple and the common mode voltage of the proposed DPWM strategy and DPWM0-DPWM3 are compared by simulation and experimental results. The results show that by the use of the proposed strategy, the values of total demand distortion of output current (I _TDD ) and weighted total harmonic distortion of output voltage (V _WTHD ) are the lowest. Furthermore, the product of switching loss and I _TDD of the proposed strategy maintains the lowest within a large range of modulation index-load angle plane. It can be concluded that the utilization of the new clamping state and optimal distribution of different clamping states could enhance the output performance of the inverter. Moreover, the high-efficiency requirement can also be realized at the same time.

In the motor drive system fed by multilevel inverter, the problems such as the complexity of reference voltage synthesis, high common mode voltage (CMV), and high switching loss are usually the obstacles to the practical application. In order to reduce the computational complexity and improve the performance of permanent magnet synchronous motor (PMSM), a novel space vector pulsewidth modulation scheme, which is practical and independent of level number is proposed for PMSM vector control. First, the space vector plane is redivided, and the coordinates of the nearest three vectors and their duty cycles can be obtained directly by determining the location of reference vector, which simplifies the calculation and does not need any iterative algorithm. Then, the optimal switching state and switching sequence are selected online so as to ensure the minimum switching loss under the premise of minimum CMV. This article provides an original method and new idea for motor drive system fed by multilevel inverter. Finally, the practicability and validity of the proposed method are verified by simulation and experiment.

The fast and stable current inner loop in the permanent-magnet synchronous motor (PMSM) control system is the key factor that ensures the torque control performance of the motor. The deadbeat model predictive control strategy can achieve fast dynamic response, but it depended on the accurate mathematical model of the motor. When the model parameter in the predictive controller has mismatched with real system, the static error or oscillations will occur in the steady state. Therefore, a novel current predictive control based on the fuzzy algorithm is proposed in this paper. The novel control strategy contained a predictive controller, a proportional-integral (PI) compensation link, a magnetic flux observer, and a fuzzy controller. According to the operation state of the motor and the model parameter mismatch of the controller, the fuzzy algorithm can adjust the effect of compensation link by weight coefficient in real time. The dynamic performance of the proposed method is guaranteed compared with the traditional deadbeat predictive current control based on the space vector pulsewidth modulation (SVPWM). When the model parameter mismatch of controller is occurred, the weight of the PI compensation link is enhanced, and the static error or oscillation of the motor system can be eliminated.

Traditional perspective considers that there is a trade-off between the postfault operational performance and the implementation complexity of the control scheme. To approach this problem, this paper proposes a natural full range minimum-loss strategy by analyzing the current relation between torque subspace and harmonic subspace. The harmonic subspace current references satisfying full range minimum-loss are generated by a simple expression according to the real-time subspace current vector features. Compared with existing fault-tolerant control methods, the proposed strategy does not need complicate fault diagnosis, as well as offline or online optimization process. The experimental results demonstrate that the proposed fault-tolerant control strategy can achieve smooth prefault to postfault transition and FRML simultaneously.