Research On Three-level Active Power Filter Based On DFT Harmonic Detection

With the continuous development of national science technology and modern industry,the power electronic equipment used in all walks of life is becoming more and more abundant,and the ensuing power grid pollution problem has to be taken seriously.Among them,the harmonic problem not only seriously affects the normal operation of precision power electronic equipment,but also threatens the stable operation of the power grid;the reactive power problem reduces the utilization rate of electrical energy,which is not conducive to the economic operation of the power grid and the dispatch of electrical energy.Therefore,the research and application of harmonic suppression and reactive power compensation technology is particularly important.Active Power Filter has received special attention due to its unique advantages over passive filters.This paper takes the T-type three-level APF as the research object,and establishes the mathematical model of the T-type three-level inverter in the three-phase static coordinate system.On the basis of analyzing the advantages and disadvantages of different filter topologies,the LCL filter is selected as the output filter,and the design range of parameter is deduced according to its performance index.The mathematical model of the LCL filter is established in the three-phase and two-phase static coordinate system respectively,which lays the foundation for the realization of the following control strategy.Aiming at the problem that the Phase Locked Loop in the time-domain harmonic detection technology is easily affected by power grid fluctuations,after briefly introducing the principle of Discrete Fourier Transform,this paper analyzes and deduces the DFT harmonic detection method in the three-phase abc coordinate system.This harmonic detection method eliminates the need for PLL design.In order to reduce the huge amount of calculation in the DFT harmonic detection method of the abc coordinate system,the process of harmonic extraction is improved,and the coordinate transformation of the signal whose harmonics need to be extracted is done,and the DFT harmonic detection method based on the synchronous rotation pq coordinate system is obtained.This method greatly reduces the computational burden of DFT and retains the same detection accuracy.Aiming at the huge optimization calculation and inefficient space vector utilization in the traditional Model Predictive Control using SVPWM,this paper proposes an improved MPC strategy that takes the output voltage of the three-level inverter as the control object to implement indirect current control.This control strategy omits the rolling optimization and value function in MPC,and can use complex and efficient multi-segment PWM technology,which greatly improves the compensation performance of APF.Although the LCL filter has excellent performance of filtering out high-frequency harmonics,it makes the mathematical model of the system more complicated,resulting in a large number of measurement demands in the prediction model for MPC,which increases the system cost.In order to solve this problem,a method of variable observation by using Linear Extended State Observer is proposed to reduce the actual measurements and save hardware resources.Due to the great model correlation of the MPC strategy,the changes of system parameters will interfere with the accurate control of the controller.The addition of LESO improves the prediction model’s resistance to disturbances caused by the changes of system parameters.In addition,the Linear Active Disturbance Rejection Control is used to replace the PI control of the traditional voltage loop,which improves the shortcomings of large overshoot and long adjustment time in PI control.The effectiveness of the above control strategies is verified in MATLAB/Simulink.Finally,the system experiment platform and its related hardware and software design are introduced,and the effectiveness of the proposed harmonic detection method and control strategy is further verified by experiments.