Research On The Control Strategy Of Three-level Active Power Filter Based On Model Prediction

In recent years,the problem of harmonic pollution in power system is becoming more and more serious,and the active power filter(APF)has attracted more and more attention because of its flexible compensation mode,strong response performance and little influence on the grid parameters.This theis focuses on the harmonic current detection method and current tracking control strategy of three-level shunt APF with three-phase three-wire NPC structure.Firstly,an improved FBD detection method based on the combination of the DDSRFSPLL and the DFT filter is presented.On one hand,the DDSRF-SPLL is used in phase locking method,in which the voltage of the power grid is decomposed to the coordinate system dq+ and dq-,and the fundamental positive sequence component is obtained by cross decoupling,which is used as the input signal of the phase locked loop to realize the accurate phase locking,in order to improve the algorithm’s adaptability in complex power grid conditions.On the other hand,the DFT filter is used to replace the traditional LPF function and the harmonic detection is realized by modifying the angular frequency,it also can ruduce the detetion delay and improve the detection accuracy.The simulation results verify the validity of the proposed method.Secondly,Aiming at the problem that the excessive number of rolling times in the traditional model predictive current method,a simplified three-level APF multi-objective model predictive current control strategy is proposed.The three level NPC structure APF mathematical model is established,and the influence of each vector on the neutral point potential is analyzed,and the number of rolling times of optimal switching vector output is reduced by sector decomposition,at the same time,making compared simulation analysis with the traditional MPCC strategy.For the control delay problem,the compensation strategy of the control delay is given.Then the switching frequency constraint is introduced into the evaluation function to realize the multi-objective optimal control,and the compensation effect is analyzed under the condition of having or not having the switch constraint,then the dynamic response performance and the adaptability to load fluctuation are verified by sudden increase of load.The simulation results show that the simplified multi-objective model predictive current control strategy has more prominent advantages of dynamic response and control flow simplification.Thirdly,a model predictive voltage control strategy based on sector classification judgment is proposed.Aiming at the problem of multiple rolling in model predictive current control method,and drawing on the idea of space vector pulse width modulation(SVPWM)based on αβ coordinate system,then the optimal switching vector can be determined directly from the 27 vectors by using the two judgment of the sector,which can leave out the link of rolling optimization and greatly improve the efficiency of search.At the same time,a qualitative control method based on small vector is introduced for the neutral point potential balance problem,then the stability relationship of the control system with delay and inductance deviation coefficient is discussed.The simulation results show that the improved predictive voltage control strategy is effective under steady state,dynamic and inductance parameters variations,and the neutral point potential balance can be well controlled.Finally,the overall structure of the three-level shunt APF with three-phase three-wire NPC structure experimental platform and the key parameters of the main circuit are designed.the improved FBD harmonic current detection method,the simplified multi-objective model predictive current control strategy and the model predictive voltage control strategy based on sector classification judgment are verified by experiments,then the operation time of different control algorithms are measured and comparatively analysised.The experimental results prove the correctness and feasibility of the research.