Control Of Three-level Photovoltaic Grid-connected Inverter

Compared with traditional two-level grid-connected inverter, three-levelgrid-connected inverter has been widely applied in high voltage and high power occasionsdue to its attractive features such as low voltage stress in switching devices, low harmoniccontents of the output current, low switching frequency and smaller electromagneticinterference. However, the control of the three-level grid-connected inverter is morecomplex due to more switching devices and more control objectives. In order to improvethe performance and efficiency of the three-level grid-connected inverter, the control systemneeds to be designed reasonable. So this paper is focus on the control technology of thethree-level photovoltaic grid-connected inverter. The main research contents are as follows:First, this paper made a contrastive analysis of the advantages and disadvantages of thetraditional three-level inverter topology and gave a T-type three-level inverter topology.Compared with the NPC-type topology, T-type topology has the advantage of low cost, lowpower consumption and can balance all the switching losses of switching devices.Mathematical model of the three-level photovoltaic grid-connected inverter is establishedbased on the T-type topology, which laid a foundation of the system control design.Second, this paper analyzed the control technology of the three-level photovoltaicgrid-connected inverter and decided to conduct an in-depth study of model predictivecontrol technology. The synthetic performance of the three-level photovoltaicgrid-connected inverter can be optimized by adjusting the weighting factors in the costfunction, so the model predictive control technology is more flexible than the traditional.Besides that, the model predictive control technology choose the optimal switching statesdirectly according to the cost function without using the modulator, so the response rate intracking the output current reference signals and balancing the neutral point voltage is veryfast. On the basis of analyzing the basic principle of model predictive control, the discretepredictive model and cost function of the three-level photovoltaic grid-connected inverterwere established. Then solutions that focus on the delay and a large amount of calculations problem existed in traditional model predictive control technology were proposed.Third, this paper analyzed the basic issues of the three-level grid-connected inverterunder unbalanced grid conditions and gave an unbalanced control technology forsuppressing the negative sequence grid current. Then an detection method for positive andnegative sequence grid voltage and a phase-locked loop technology based on the decoupleddouble reference frame were proposed.Which were aimed at the issues that positive andnegative sequence grid voltage need to be detected and traditional phase-locked loop can’tlock grid voltage phase quickly and accurately under unbalanced grid conditions. Theproposed methods can detect the positive and negative sequence grid voltage and lock gridvoltage phase quickly and accurately.Fourth, this paper built a system simulation model based on Matlab/Simulink andconducted the simulation of the model predictive control for the three-level photovoltaicgrid-connected inverter. Simulation results show that the designed model predictivecontroller can well tracking the reference current, balancing the neutral point potential andcontrolling the output current balanced even under unbalanced grid conditions. Thedynamic and static performance of the system is very well.Fifth, on the basis of simulation, the hardware and software system of the three-levelphotovoltaic grid-connected inverter were designed. Then a30kW three-level photovoltaicgrid-connected inverter experimental circuit was built and the validity of the proposedmodel predictive control technology was verified. Experimental results show that thedesigned model predictive controller can well control grid current and balance the neutralpoint potential. Besides that the total harmonic distortion is very low, the system powerfactor is very high, which meets the system design requirements.