Current Prediction Control Of Permanent Magnet Synchronous Machine Based On Lumped Parameter Model

The permanent magnet synchronous machine(PMSM)has superior advantages such as high power factor,high efficiency and low temperature rise,so it is widely used in electric vehicles,CNC machine tools and other fields.In a PMSM control system,the dynamic performance of current loop has a direct influence on the stability of machine control,while the conventional current loop PI control strategy gradually fails to meet the increasing practical demands in terms of dynamic performance and robustness.The predictive control has advantages such as predictability,fast dynamic response and easy for implementation.Therefore,both current industry and academia start to widely study and employ it to replace the conventional PI control for current loop.In this thesis,the predictive control model of PMSM is derived,and a deep analysis of the principle and performance of deadbeat current predictive control(DCPC)is conducted afterwards.Meanwhile,the influence of inverter nonlinearities and machine parameter uncertainties on DCPC is also studied,and relevant solutions are studied and given as follows:(1)In order to compensate the inverter nonlinearities,a compensation strategy based on the model reference adaptive system(MRAS)algorithm is proposed in this paper,which is based on the fundamental model of PMSM.Firstly,the error voltage due to inverter nonlinearity is observed online,and the observed result will be directly used for the compensation of reference voltage,so as to realize the online compensation of disturbance voltage due to inverter nonlinearity.The method is simple for debugging and has a good compensation effect.However,its performance relies on accurate machine parameter information.In view of this problem,a compensation strategy for inverter nonlinearity based on adaline neural network(Adaline NN)is further developed.It is based on the observation of the amplitude of6 th harmonic component generated by the disturbance voltage due to inverter nonlinearity in the dq-axis reference frame,which will be minimized through an online iterative optimization scheme.Finally,the compensation of inverter nonlinearity will be achieved.Compared with the previous MRAS scheme,this method does not depend on the machine parameters and can achieve a high accurate online compensation.(2)In order to further improve the robustness and s teady-state accuracy of PMSM control system,this paper studies control strategies which combines the compensation strategybased on MRAS with DCPC and combines the compensation strategybased on Adaline NN with DCPC.Compared with the traditional DCPC,the DCPC strategy based on inverter nonlinearities compensation can effectively reduce the dq current fluctuation and torque ripple,and improve the PMSM control performance.(3)The accuracy of the parameter model of PMSM predictive control often directly affects the performance of DCPC algorithm.Therefore,the DCPC algorithm has a high dependence on machine parameters,especially the inductance and permanent magnet flux.The sensitivity of conventional DCPC algorithm to machine parameter disturbance has been analyzed,and it is found that inaccurate inductance error and permanent magnet flux will result in a steady state observation error of currents,and a large error in inductance value can even result in a divergence in current observation.In order to reduce the influence of mismatched machine parameters and inverter nonlinearity on the control performance of DCPC,an improved DCPC model based on the lumped parameter model has been proposed in this thesis.The disturbance voltage caused by mismatched parameters and voltage disturbance due to inverter nonlinearity are represented as a lumped parameter,and an adaptive sliding mode observer is designed for observation and aiding the prediction of currents.Finally,the observed lumped parameters ar e used for the compensation of the output voltage so as to improve robustness of DCPC.The above two control strategies have been verified through both simulations and experiments,and all relevant key issues have been thoroughly analyzed with corresponding solutions.Compared with the conventional scheme,the lumped parameter model based DCPC can be an effective scheme to eliminate the influence of inverter nonlinearity,and also can be free of the dependence of machine parameters.In this case,it has a good theoretical significance and great values in engineering applications.