Research On Model Predictive Control Strategy Of Three-phase Current Source PWM Rectifier

The current source PWM rectifier(CSR)has the characteristics of wide-range stepdown output,high reliability of DC short-circuit protection,and easy parallel operation.It can be widely used in DC microgrids,data center front-end rectifier interfaces and multielectric aircraft and other fields.Therefore,optimizing the control strategy to improve the system stability is of great significance to the rapid development of CSR.Among many control strategies,Model Predictive Control(MPC)has the advantages of intuitive modeling,fast dynamic response,and easy handling of system constraints,and has gradually become the main research direction in the field of power electronic system control.This paper takes CSR as the research object,and analyzes and studies its model predictive control strategy.Firstly,according to the analysis of CSR topology and working mode,the mathematical models of CSR in different coordinate systems are deduced from Kirchhoff’s voltage-current law.Secondly,according to the basic principles and characteristics of finite set model predictive control,a state space prediction model of CSR is established.A large number of operations in the prediction process will cause the controller to delay the action.After the delay compensation is performed by the two-step method,the corresponding cost function is established according to the control objective of the CSR.The traditional model predictive control ignores the LC resonance on the CSR grid side.In order to suppress the resonance,the effectiveness of the passive damping control scheme is analyzed.Then,some improvement strategies are proposed for the shortcomings of traditional model predictive control schemes.In order to realize the stable tracking reference value of the CSR DC output,in the traditional model predictive control,the DC side adopts the classic PI control,and there is PI parameter tuning.To avoid complex PI parameter tuning,a hybrid model predictive control scheme is proposed.The scheme adopts Finite Control Set Model Predictive Control(FCS-MPC)on the AC side of the CSR and Deadbeat Predictive Control(DBPC)on the DC side.The source of the program calculation burden of traditional model predictive control is analyzed,and a prediction scheme that directly uses the input current vector to establish the cost function is proposed,which reduces the program calculation burden.At the same time,the switching frequency constraint term is added to the cost function to reduce the switching frequency and reduce the switching loss.Adopting passive damping control scheme will increase the power consumption of the system.Therefore,an active damping scheme suitable for hybrid model predictive control is proposed.The scheme uses the input current vector calculated by the CSR grid-side current and the AC-side capacitor voltage to establish a cost function,which suppresses the LC-side LC resonance of the CSR and improves the power conversion efficiency.A state observer is designed to observe the capacitor voltage on the AC side of the CSR in real time,reducing the use of voltage sensors and system design costs.Finally,a system simulation model is built on the Matlab/Simulink platform to verify the feasibility of the proposed control strategy.An experimental prototype of a three-phase current source PWM rectifier is built to verify the proposed control strategy.The experimental results show that the proposed control strategy has good steady-state and dynamic performance under the condition of grid balance;in the process of grid voltage frequency conversion,the proposed control strategy still has good dynamic performance.