The multiphase machines have been widely explored in industries and academia in past years, since they can achieve the advantages of lower power rating per phase, lower torque pulsation and better fault tolerant capability. Thus, the multiphase machines are more preferred than the three-phase counterparts in high power and safety-critical applications, such as electric ship propulsion, elevator, wind power generation and so on. In addition, the model predictive control (MPC) has emerged as a competitive technique for the control of machine drives and power electronics in recent years. When compared with the conventional vector control and direct torque control (DTC), the MPC accommodates the advantages of fast dynamic response, intuitive implementation and easy inclusion of system nonlinearities. Despite of all the endeavors, there are still many challenges imposed on the MPC for multiphase machines. Thus, in this thesis, the advanced control strategies based on MPC for the asymmetrical six-phase permanent magnet synchronous machines (PMSMs) are explored in depth.

The limitation and the problems of the conventional MPC when it is applied in conjunction with the six-phase machine are briefly described here. First, in the conventional MPC, the number of prediction vectors increases exponentially as the number of phase increases. So, compared with the three-phase machine drive, much larger computation time is required in the MPC for six-phase PMSM machine. Secondly, high amount of harmonic currents is introduced due to the small leakage impedance of the winding structure of asymmetrical six-phase machines. The harmonic currents do not contribute to the torque production. Instead, copper losses will be introduced by the harmonic currents, which spoils the system efficiency. Thirdly, there is always a weighting factor involved in the model predictive torque control (MPTC) to achieve the simultaneous control of the torque and stator flux. Unfortunately, the tuning work of the weighting factor is a tedious work based on trial and error procedures due to the lack of theory support. Fourthly, the torque command tracking capability will be deficient if all the applied voltage vectors share the same magnitude, which will cause large torque ripple. Fifthly, the fault-tolerant control based on MPC for multiphase machine is still not well investigated. In addition, all the existing literatures adopt the reduced-dimension mathematical model of the machine, where the transformation matrix and controller structure are reconfigured under the post-fault operation.

Thus, the purpose of this thesis is to solve the aforementioned problems involved in the MPC for six-phase PMSM machines. First, the research background, research objectives and thesis outline are introduced in Chapter 1. Next, Chapter 2 is dedicated to the survey of all the control strategies for electric machines. In Chapter 3, the mathematical modelling of the six-phase machine, including the magnetomotive force (MMF) analysis, double d-q modelling and the vector space decomposition (VSD) modelling are investigated. In Chapter 4, a simplified MPTC for the six-phase machine is proposed, where the current harmonics suppression and computation time reduction are achieved simultaneously by using a two-step look-up table. In Chapter 5, the deadbeat current control is employed to obtain a reference voltage vector and the cost function is used to select the optimal vectors that are closest to the reference voltage vector directly. Moreover, the synthesized vectors are employed to achieve zero harmonic voltage theoretically. In Chapter 6, to simplify the predictive model and lower the control complexity, a flux constrained model predictive control is proposed. A comprehensive comparison in terms of computation time, predictive model and machine performance among the proposed method and other benchmark methods is presented. Chapter 7 is devoted to the remedial strategy for the single-phase open-circuit fault of the six-phase motor, which is based on the normal transformation without the controller reconfiguration by introducing a perturbation term. Finally, the conclusion and the future prospects are drawn in Chapter 8.