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

With the development of new energy grid-connected power generation(for example,photovoltaic and wind power),the application of high-power grid-connected inverters has increasingly become an important part.In these high-power grid-connected inverter applications,their inverters often use three-level topology.Conventional control strategies are difficult to achieve multiple goals such as current control,midpoint potential balance,and lowest switching loss of high-power grid-connected inverters.optimized control.At present,the application of Model Predictive Control(MPC)in high-power grid-connected inverters has received more and more attention.However,the model predictive control of high-power three-level grid-connected inverters still needs to overcome several key technical issues: such as harmonic suppression,weight factor selection,parameter robustness,and economic operation.In this regard,this article takes the three-level grid-connected inverter as the research object to study the above problems.The main research work and innovations are as follows:1)Aiming at the current harmonic suppression problem of high-power three-level grid-connected inverter model predictive control,two harmonic suppression strategies of model predictive control are proposed.One is based on Adaptive Selective Harmonic Elimination(ASHE)algorithm model predictive control,ASHE algorithm is used to eliminate specific harmonics in grid-connected current;the other is model predictive control based on the harmonic suppression algorithm of trigonometric function orthogonality,by suppressing the harmonics A target is embedded in the cost function of model predictive control to achieve harmonic suppression of grid-connected current.2)Aiming at the problem of selecting weighting factors for model predictive control of high-power three-level grid-connected inverters,the reasons why it is difficult to select weighting factors for model predictive control in multi-objective optimization are analyzed,and a weightless factor model based on hierarchical control is proposed.Predictive control solves the problem of difficult selection of weighting factors.3)Aiming at the parameter robustness problem of the model predictive control of the high-power three-level grid-connected inverter,a predictive model of the three-level grid-connected inverter system is established.On this basis,two robust model predictive controls are proposed.One is a non-parameter model predictive control strategy based on a fixed window optimization algorithm.This control strategy does not require the use of system parameters,thereby improving the parameter robustness of model predictive control;the other is based on adaptive linear neural network(Adaptive linear neural network).Linear Neuron,ADALINE)algorithm model predictive control,ADALINE algorithm is used to update parameters online,to achieve robust model predictive control.4)Aiming at the economic operation problem of the model predictive control of the high-power three-level grid-connected inverter,in order to reduce the use of sensors,a robust model predictive control strategy without voltage sensors is proposed.The voltage is estimated by the extended state observer.The voltage sensor is omitted,and robust model predictive control is realized through the ADALINE algorithm.The proposed method further reduces the operating cost of the grid-connected inverter and improves the reliability of the system.In addition,an experimental platform for a three-level grid-connected inverter was built to verify the proposed control strategy.