Research On Predictive Control Of Three-Phase Three-Wire Active Power Filter

A large number of non-linear loads make the power grids increasingly polluted by harmonics,causing attention and attention of relevant departments.As an effective means of dynamically suppressing harmonics,active power filter(APF)has gradually become a research hotspot.In this paper,the three-phase three-wire parallel APF is taken as the research object,and its harmonic detection,harmonic tracking and model prediction algorithms are analyzed separately.Firstly,this paper introduces a specified subharmonic detection algorithm based on semi-periodic DFT.This algorithm is improved on the basis of full-period discrete Fourier,and the calculation time is saved by half.At the same time,based on the halfcycle DFT.Combined with the sliding window iterative algorithm,the calculation time is shortened in one sampling period,and the computational efficiency is greatly improved.In the aspect of specifying subharmonic tracking,the working principle and control effect of proportional resonance(PR)and vector resonance(VR)controllers are introduced.According to the topological structure of the three-phase three-wire APF,the discrete mathematical model and the midpoint potential balance mechanism are analyzed in detail,and the influence of the neutral potential balance on the neutral point is discussed.This paper analyzes the method of constructing the value function by rolling out the optimization of all switching vectors to select the optimal switching state in the traditional model predictive current control strategy,and introduces a multi-step prediction based on the low control precision of the traditional algorithm.The current control strategy,which uses a multi-step vector optimization algorithm to find the most suitable switching state to act on the converter,improves the system control accuracy.Aiming at the shortcomings of the model predicting the large amount of computation in the current control strategy,a voltage control strategy is introduced,which is the main target of voltage tracking as a value function.It only needs two levels of sector judgment to directly locate the optimal switching state.It eliminates the complicated operation process of scrolling and optimizing,which greatly improves the efficiency of optimization.At the same time,using the effect of small vector on the midpoint potential,a midpoint potential balance method based on redundant small vector selection is proposed.Combined with the voltage model prediction algorithm,the midpoint potential is effectively controlled.The hardware and software of the three-phase three-wire APF experimental platform are designed.The experimental platform is used to verify the harmonic detection algorithm based on semi-period DFT and the model prediction based on twostage sector prediction.The correctness of the voltage control algorithm is also tested for the midpoint potential balance control algorithm based on redundant small vector selection.