Model Predictive Current Control For Dual-Three Phase PMSM With Low Inductance

Different from traditional multi-phase machine,low inductance multi-phase machine possesses a faster current variation rate due to the weakened current suppression ability.Combining the finite-control-set model predictive control(FCSMPC)strategy with low inductance machine could further exert the advantage of quick response speed.However,the too-fast current change caused by low inductance requires a higher control frequency that cannot by increased arbitrarily due to the limitation of hardware conditions,which seriously affects the control performance.Therefore,how to combine low inductance motor with FCS-MPC at the existing frequency is the main problem to be solved.In order to solve the above problems,this thesis takes the low inductance dual-three-phase permanent magnet synchronous machine(DTP-PMSM)as the research object,and studies the optimization algorithem of FCS-MPC for low inductance multi-phase machine system.Firstly,through the mathematical model of DTP-PMSM based on space vector decomposition,the implementation process of FCS-MPC on dual three permanent magnet synchronous machine is researched.Furthermore,this thesis reveals the influence of delay effect on performance and proposes a delay compensation strategy.The feasibility of the method is verified by simulation and the influence of different inductance values on control performance is analyzed.In the meantime,a multi-phase machine drive platform is established based on DSP28335 and the traditional predictive control scheme is verified by experiments.Secondly,aiming at the high complexity of traditional scheme and the rapid current variation caused by low inductance,a virtual vector MPCscheme with duty cycle optimization is proposed.Through the synthesis of virtual voltage vectors,the predictive model is simplified while suppressing the current harmonics.Hence,the calculation amount is largely reduced and the weighting factor design is also omitted.The duty cycle is also introduced to adjust the action time of voltage vectors in each period,which effectively address the issues of low inductance.The feasibility of proposed scheme is verified by simulated and experimental results.Finally,in order to further improve the steady-state control accuracy,a multi virtual vector model predictive current control is proposed.Two effective virtual vectors and one zero vector are selected to work together in each period.The action time of three vectors is designed according to deadbaeat principle so as to ensure that the current predictive values can accurately reach the reference values.Moreover,the output range of composite vector is analyzed and the three different schemes are compared.The effectiveness of the multi virtual vector scheme are verified through simulation and experiment,and its existing problems are also discussed.