Application Of Model Predictive Control In LCL-type Four-leg Active Power Filter

Nowadays,power electronic devices are widely used in various fields of people’s production and life with the continuous development and improvement of power electronic technology.The problems of harmonics and three-phase imbalance in low-voltage distribution networks are becoming more and more serious.The four-leg active power filter(APF),as a device that can comprehensively solve the problems of harmonics and three-phase unbalance,has received widespread attention.Output PWM voltage of APF contain abundant switching harmonics.In order to suppress the switching harmonics in the compensation current,so output filter is required.The LCL filter has received more and more attention with excellent high-frequency attenuation ability.This thesis takes LCL-ype two-level four-leg APF as the research object,and studies the resonance suppression strategy of LCL and the application of model predictive control in LCL-type four-leg APF.Aiming at the coupling problem of mathematical model of LCL four-leg APF in abc static coordinate system,the zero sequence model and non-zero sequence model are separated by coordinate transformation to αβγ coordinate system.The mathematical decoupling is realized.The phase structures in αβγ coordinate system are independent of each other,which facilitates the design of LCL parameters and subsequent control algorithm.By analyzing the influence of each parameter of LCL filter on its performance and the constraints of parameter design,the parameters of four branch LCL filter are designed by graphic method.Aiming at the accuracy problem of harmonic current prediction algorithm in APF model prediction algorithm,a harmonic current prediction algorithm is proposed.The simulation results show that the improved repeated prediction algorithm not only has high steady-state prediction accuracy,but also has fast dynamic prediction speed for harmonic command current.In order to solve the resonance problem of LCL filter,the active damping method of capacitance current proportional feedback is adopted.Combined with simplified 3D-SVPWM,a model predictive composite control scheme based on Deadbeat theory is proposed,which is applied to the LCL-type four-leg APF.The effectiveness of the composite control scheme is verified by simulation.For traditional FCS-MPC,the limitations of active damping based on state feedback are analyzed.Useing the characteristics of multi-objective control in model predictive control,study the method of LCL resonance suppression in model predictive control.Based on the discrete mathematical model of LCL-type four-leg APF,the distribution of its voltage vector is analyzed,and the principle of the FCS-MPC algorithm based on the real vector is analyzed.In order to improve the control accuracy of APF and the effect of harmonic suppression,the discrete vector modulation method is used to introduce virtual vector to increase the available voltage vector,and the synthesis method of four leg virtual vector is studied.In order to reduce the computation of traversal optimization,the process of rolling optimization algorithm is optimized.The influence of delay on the system is studied,and the delay compensation strategy of one beat ahead is adopted to compensate the delay.The simulation results show that FCS-MPC optimization algorithm based on the virtual vector has better resonance suppression effect and APF compensation accuracy than FCS-MPC algorithm based on the real vector.The simulation results show that the resonance suppression effect of LCL and the compensation accuracy of APF can be effectively improved by properly increasing the weight coefficient of capacitor voltage and inverter side current.Finally,an experimental platform of LCL-type four-leg APF was established based on Typhoon HIL.The designed LCL parameters and the above model predictive control strategy were verified through experiments,and the experimental results were consistent with theoretical analysis and simulation.