Research On Current Predictive Control Strategy For Permanent Magnet Synchronous Motor

The Permanent Magnet Synchronous Motors(PMSMs)have high power density and strong overload capability,making them widely used in real-life applications.Model Predictive Control(MPC)enables efficient control of PMSM systems.However,traditional MPC control strategies have issues such as sensitivity to model variations and parameter mismatches,leading to suboptimal control system performance.In this thesis,focusing on PMSMs,the main research objective is to propose optimal dualvector combination MPC control strategy and an improved MPC control strategy with current prediction error compensation method,based on the foundation of model predictive control.These strategies aim to improve the steady-state performance of the control system.Firstly,according to the mathematical model of the Permanent Magnet Synchronous Motor(PMSM),different coordinate system models of the PMSM are derived through coordinate transformation.The motor prediction model is obtained by discrete methods,and the cost function is determined.The error caused by the delay problem is analyzed,and the delay problem is compensated through a two-step prediction method.Traditional Duty Cycle Model Predictive Control(Duty-MPC)has limitations in voltage vector selection,where the second voltage vector can only be a zero vector,leading to suboptimal performance.To address this problem,this thesis proposes an MPC control strategy based on optimal dual-vector combination.This method considers all possible combinations of voltage vectors and calculates the corresponding duty cycle.Firstly,the predicted current values are transformed to obtain the reference voltage vector.The optimal voltage vector is selected through reference vector transformation,sector transformation,and optimal vector selection.This approach reduces computational complexity while improving system performance.The average back electromotive force(EMF)within three cycles is computed to reduce calculation errors.MATLAB/Simulink simulation is set up to compare and verify the control effects of duty cycle MPC and the optimal dual-vector combination MPC.The simulation demonstrates that the proposed control strategy exhibits good steady-state characteristics,validating the effectiveness of the control strategy.Furthermore,a thorough analysis and study are conducted on the causes of prediction errors in Model Predictive Control(MPC).The formula for prediction errors is derived,and traditional current prediction error compensation methods are introduced based on the varying impact of different motor parameters.Building upon this method,it is combined with the optimal dual-vector MPC control strategy to optimize the traditional approach.Firstly,the space voltage vector space is divided into regions,and the region to which the applied voltage vector belongs is determined.Simultaneously,the prediction error between the predicted current value and the actual value is calculated.Based on the different voltage vector regions,the predicted current value is compensated.After compensation,the current undergoes optimal dual-vector combination selection,resulting in the selection of the optimal dual-vector combination.This method compensates for the current prediction error,thereby improving the control accuracy and robustness of the PMSM control system.MATLAB/Simulink simulation is set up to verify the compensation effect of the improved current prediction error compensation.The simulation demonstrates that this control strategy has certain compensation effects and achieves higher control accuracy.The proposed control strategies are experimentally validated using a PMSM experimental platform.A comparative analysis is conducted to evaluate the control effects of different control strategies and the traditional control strategy.The experiments demonstrate that both of the proposed control strategies in this thesis exhibit good control performance and possess strong system robustness.This thesis includes 61 figures,5 tables,and 82 references.