Research On Model Predictive Control Of Three-level PWM Rectifiers

With the rapid development of power electronics technology,PWM rectifiers,as converters at grid interfaces,are widely used in photovoltaics,motor drives,and charging piles.This thesis takes the three-level NPC PWM rectifier as the research object,and studies its modulation strategy and control strategy.First of all,this thesis introduces the topology and operating principle of the threelevel NPC PWM rectifier in detail.On this basis,the current mathematical model is deduced and established under three-phase static coordinate system,two-phase static coordinate system and two-phase synchronous rotating coordinate system respectively.And the power mathematical model is established based on the instantaneous power theory.Secondly,the deadbeat predictive power control is studied.Two-step prediction is used to eliminate the influence of system delay on control performance.Aiming at the problem of power steady-state error in two-step prediction,a quasi-integral feedback observer is studied to correct the estimated value of power at the next moment,and the steady-state no-static-error control of active and reactive power is realized.Thirdly,aiming at the slow dynamic response of the voltage outer loop PI controller and the large fluctuation of the DC bus voltage,the design idea of dynamic given reference is introduced,and the nonlinear disturbance observer is used to replace the DC side current sensor to realize real-time observation of the load power.In addition,the loss power of the parasitic resistance of the inductor is also regarded as power disturbance,which reduces the dependence of the dynamic given link on the accuracy of the system parameters and improves the robustness of the system.Fourthly,the problem of large power fluctuation in traditional model predictive direct power control is analyzed,and the suppression effects of two two-vector-based model predictive direct power control methods on power fluctuation are studied.Among them,the two-vector-based model predictive control based on optimal vector optimization adds a voltage vector to adjust the error of active power under the action of the optimal vector on the basis of the traditional model to predict the optimal vector,and suppress power fluctuation.However,the two-vector-based model predictive control strategy based on optimal vector optimization has a conservative problem in vector selection,which leads to six large power fluctuations in one fundamental wave cycle.The two-vector-based model predictive control based on sector judgment cancels the restriction on the first voltage vector,judges the sector where the reference voltage vector is located based on deadbeat control,optimizes the two vectors in the sector at the same time,and selects the optimal vector combination.Finally,the three-level PWM rectifier experimental platform is introduced in detail,and the neutral-point balance control and model predictive direct power control algorithms are verified.