Research On Model Predictive Control Method Of Three-level Grid-connected Inverter

As the core device to realize AC/DC conversion of electric energy,grid connected inverter take part in distributed generation system,electric vehicle and other new energy fields.Neutral Point Clamped three-level inverter has been applied because of its advantages of low harmonic content and low withstand voltage of switching devices,and so on.But it brings more control problems,including current tracking control,DC side midpoint voltage balance,common mode voltage and switching frequency suppression.The finite control set model predictive control(FCS-MPC)algorithm is widely used in the control of three-level inverters because of its simple control principle,no need for PWM modulators,and easy handling of multiple targets.There are still a lot of questions in this algorithm yet.This thesis regards NPC three-level inverter as the research object and adopts FCS-MPC to study it.In view of the problems of FCS-MPC algorithm in practical application,such as control delay,large amount of calculation and mismatching of inductance parameters,the research was conducted.On this basis,the multi-objective model predictive control method of three-level inverter was studied.The key study contents were as follows.Firstly,the three level modes of NPC three-level inverter are studied.Based on the basic requirements of FCS-MPC algorithm,the discrete prediction model and objective function of network side current and dc side midpoint voltage are established,and the control flow of FCS-MPC algorithm on the inverter is given.Then,the effect of sampling and calculation on the control of the system is learned,and the delay compensation is made by the forward prediction.Aiming at the problem of large amount of calculation in the model predictive control,the space vector area division idea is used to divide the entire space voltage vector diagram into six areas.The optimal region is obtained by the Lyapunov function based on the current tracking error,reducing the 27 candidate switch states to 10,and the number of algorithm prediction and optimization is reduced.Next,the effect of the inductance parameter mismatch on the current control performance is studied.In line accordance with the current relationship,a self-correction method of the inductance parameter is proposed to correct the model inductance in real time to avoid the inductance error affecting the control effect.On the basis of optimizing the FCS-MPC algorithm,for the four control objectives of the three-level inverter: grid-side current,midpoint voltage,common-mode voltage and switching frequency,the model predictive control method is used to carry out multi-objective comprehensive control.The linear weighted multi-objective control strategy based on three-scale analytic hierarchy process and the weightless factor model predictive control method based on hierarchical sequence multi-objective optimization are designed respectively.While meeting the control requirements of current and neutral point voltage,the common mode voltage and average switching frequency are reduced,thus the comprehensive control of the system is implemented.Finally,the physical system of three-level inverter is established to certify the effectiveness of model predictive control method mentioned in this paper.