Research On Model Predictive Control Of Five-Phase Flux-Switching Permanent Magnet Motor Based On Disturbance Observation

Compared with three-phase motor drive system,five-phase motor drive system has the advantages of achieving high power output in relative low voltage,more control degrees of freedom and satisfactory ability in fault tolerance.As a special structure of five-phase permanent magnet synchronous motor(PMSM),five-phase flux-switching permanent magnet(FSPM)motor has more advantages in power density and heat dissipation design.In this thesis,a five-phase stator-20-slots/rotor-18-poles FSPM motor is applied as the experimental object to study the five-phase FSPM motor model predictive control(Model Predictive Control,MPC)anti-disturbance and relative issues.The key works are summarized as follows:1.The structural characteristics and basic working principles of the five-phase FSPM motor are described systematically,then the mathematical models of the five-phase FSPM motor in different coordinate systems are constructed.To simplify the control of harmonic subspace,a modulation strategy based on virtual voltage vector is proposed.2.Two five-phase FSPM motor model predictive current control(MPCC)methods are introduced,namely the conventional MPCC method and the deadbeat model predictive current control(Deadbeat MPCC,DB-MPCC)method,and comparative studies between them are fully conducted.3.The influence of parameter mismatches on the conventional MPCC and the DB-MPCC methods are analyzed separately.Moreover,the system stability under parameter deviations is also theoretically analyzed.4.To eliminate the static current error caused by parameter mismatches,two improved DBMPCC methods are proposed based on the integral method and the Luenberger observer separately.The basic principles and implementation methods of these two methods are introduced in detail,and a specific and practical observer gain design scheme is proposed.The comparative study of steadystate performance under different parameter deviations and dynamic performance of these two improved DB-MPCC methods are well conducted.5.A five-phase FSPM prototype experimental platform is built,and the related algorithm codes are programmed and compiled.Sufficient experimental researches are conducted to verify the effectiveness of the proposed algorithms.