Research On Model Predictive Control For Quasi-Z-Source Inverter Without Weighting Factors

The quasi-Z-source inverter(q ZSI),as a new inverter topology,has the merits of continuous input current,bridge arms short-circuit immunity,boost and buck functionalities,and has a broad prospect in electric vehicles,photovoltaic,wind power and other fields.Meanwhile,model predictive control(MPC)has become an effective control algorithm for quasi-Z-source inverter due to its advantages of simple principle,fast dynamic response and easy integration of multiple control variables.In the traditional model predictive control of quasi-z-source inverter,the cost function usually contains weighting factors.However,there has been no thorough and systemic design strategy and theoretical system of the weighting factors so far.Moreover,the most commonly used weighting factors adjustment strategy is trial-and-error method,which is time-consuming and sophisticated,and has become a key problem hindering the further development of model predictive control.Therefore,the quasi-z-source inverter is regarded as the research object,and the model predictive control strategy is further studied in this paper.Firstly,the working principle and mathematical model of the quasi-z-source inverter are analyzed and established,and the model predictive control strategy is introduced in detail from the aspect of the basic principle,cost function and weighting factors design,delay compensation and future reference value prediction strategy.Secondly,from the perspective of basic principle and design strategy,the model predictive control strategy without weighting factors for the quasi-z-source inverter proposed in this paper is analyzed concretely.In terms of inductor current control,a modified inductor current reference value is derived.By comparing the relationship between the measured value of inductor current and the modified reference value of inductor current,whether to choose the shoot-through state as the optimal switching state can be determined,thus the weighting factor of inductor current term is avoided.In terms of capacitor voltage and output current control,the proposed strategy constructs two separate cost functions for the capacitor voltage and output current,and then rationally controls the capacitor voltage and output current according to different voltage vector selection mechanisms,avoiding the weighting factors of capacitor voltage and output current terms.In addition,a load parameter adaptive module is constructed for the load parameter perturbation problem,and the adaptive laws of lumped electromotive force and lumped inductance are derived.As a result,the prediction model can be updated in real time according to the adaptive laws to improve the robustness of the system.Finally,the simulation model is establiashed in MATLAB/SIMULINK simulation software,and the experimental platform is built based on the real-time digital control system RTU-BOX,and the traditional model predictive control strategy and the proposed strategy in this paper are analyzed in depth.The results have demonstrated that the proposed strategy eliminates the weighting factors and makes the control variables better controlled,and owns the merits of good output harmonic performance,anti-disturbance ability,robustness and less computation compared to the traditional model predictive control strategy.It is noteworthy that the proposed control strategy has certain universality,which is not only applicable to different topological circuits,but also applicable to situations with multiple constraints.It can provide research ideas and reference significance for other topological circuits and control strategies.