Research On Control Strategy Of PWM Rectifier Based On Model Prediction

With the development of new energy technology,the PWM rectifier is an important interface between power equipment and the power grid,and its core control goal is achieving effective control of the AC side and the DC side.The three-level PWM rectifier has the advantages of high withstand voltage,low harmonic content and simple control scheme.At present,it is widely used in various production and life fields.This article takes the NPC type three-level PWM rectifier as the research object,and deeply researches the synchronous detection technology and control strategy based on the PWM rectifier.This thesis has carried out theoretical analysis and simulation verification on the traditional synchronization signal detection method.Aiming at the problem that SRFPLL cannot accurately lock phase under the condition of unbalanced grid voltage,a grid voltage synchronization technology based on SDFT-PLL is proposed.The algorithm can effectively eliminate DC bias and high-frequency harmonics through the SDFT module,and achieve precise phase lock within one cycle under the conditions of grid distortion and unbalance.The SDFT-PLL algorithm is simulated and analyzed by MATLAB/Simulink.This article derives the model of the NPC type three-level PWM rectifier,and thesis establishes the corresponding mathematical model.On this basis,the working principle of FCS-MPC is introduced.In the traditional three-level FCS-MPC,its value function is optimized to expand the control target from single to multiple.In addition to tracking the current on the AC side,the midpoint potential control and switching limit are also added.Through online scrolling optimization and feedback self-correction,the influence of dynamic fluctuations of the controlled object can be effectively overcome,and effective control of the PWM rectifier can be realized.The traditional three-level FCS-MPC is simulated and analyzed by MATLAB/Simulink.This thesis studies the multi-step model predictive control strategy of the NPC type three-level PWM rectifier,and proposes a model predictive control strategy based on an improved sphere decoding algorithm.On the basis of multi-objective model predictive control,using the idea of spherical decoding for reference,the value function is matrix optimized through iterative backtracking.This thesis solves the problem of high computational complexity in multi-step predictive control by means of equivalent conversion midpoint potential control and switching sequence limit constraints.In view of the problems in the implementation of the sphere decoding algorithm: I.In the transient process,the hypersphere radius r(k)of the sphere decoding algorithm may be too large,which causes the amount of calculation to increase sharply.II.The initial radius r(k)of sphere decoding is too small,resulting in no complete switching sequence inside the hypersphere.III.Low switching frequency leads to increase in AC side harmonic THD.This article proposes several improvement measures: Aiming at problem I,this article restricts the number of complete switch sequences inside Hypersphere;Aiming at problem II,this article selects the predicted switch sequence generated at the previous moment to correct the switch state;Aiming at problem III,this thesis chooses a harmonic preprocessing strategy based on grid-side current feedback.Through MATLAB/Simulink simulation,the effectiveness of the proposed algorithm is analyzed.Finally,based on the NPC type three-level PWM rectifier experimental platform,the experimental verification of related algorithms is completed.There are 68 figures,10 tables and 82 references in this thesis.