Study On Model Predictive Control Strategy Of Three-level PWM Rectifier

With the development of industry,railway and new energy resources,the application of PWM rectifiers is becoming more and more extensive.Thus,it is quite necessary to study high-performance PWM rectification technology.Taking NPC threelevel PWM rectifier as the research object,the phase-locked loop and control strategy are thoroughly studied in this thesis.The main work and conclusions are as follows:Traditional synchronous reference frame phase locked loop(SRF-PLL)has problems of complicated PI parameters setting and poor phase lock accuracy in the case of grid voltage imbalance.This thesis briefly introduces the phase-locking principle of SRF-PLL technology,and analyzes the influence mechanism of negative sequence components on the phase-locking accuracy of SRF-PLL technology when grid voltage is unbalanced.Based on that discussed above,a SOGI-FGF phase-locked algorithm,which can precisely lock the phase in the case of grid voltage imbalance by adjusting only one parameter,is proposed according to the principle of second-order generalized integrator(SOGI)and fixed gain filter(FGF).In addition,a thorough introduction and stability analyzation are made for the algorithm.Finally,the SRF-PLL and SOGI-FGF phase-locked methods are simulated in Matlab/Simulink to make a comparison.To avoid the problem of high voltage amplitude jump on the AC side while traditional model predictive control(MPC)is adopted in three-level PWM rectifier,a constraint is utilized to limit the direct switching between the switch states 1 and-1.First,the mathematical model,starting from the basic principle of PWM rectifier,is derived in detail to pave the way for MPC algorithm of PWM rectifier.Second,the MPC principle is introduced briefly,and the prediction model of three-level PWM rectifier is obtained by combining the mathematical model.The AC current tracking performance,neutral point potential balance and switching times are considered as the value functions,and the optimal switching state is obtained by rolling optimization to realize effective control of the PWM rectifier.A multi-step MPC control strategy based on sphere decoding is proposed to solve the problems of local optimization in single step MPC algorithm and excessive computation in multi-step MPC algorithm.First,the basic principle of spherical decoding algorithm is introduced.In addition,the value function is converted into the least squares problem that can be solved by spherical decoding algorithm,and then the multi-step MPC is optimized by using the branch and bound idea.Considering the prediction formula of the midpoint potential cannot be directly added to the conversion process of value function,the midpoint potential is taken as a constraint to limit the update of super sphere radius in spherical decoding algorithm,so as to control the midpoint potential balance.In the end,the control method described in this thesis is verified on the experimental platform of NPC three-level PWM rectifier.There are 69 figures,7 tables and 83 references in this thesis.