Research On Model Predictive Control Of Grid-connected Converters

Today, the generation of electrical energy is largely based on fossil fuels and nuclear power in the world. But the use of non-renewable energy has been proven to be one of the main causes of environmental problems. So it is quite necessary to increase the penetration level of renewable energy and achieve sustainable development, in response to the rapid increasing demand for energy and environmental protection. Currently, the growth of renewable electricity has exceeded that of non-renewable electricity. Wind and solar energy are the most popular options for renewable electricity. As the connector between the renewable energy and grid, grid-connected converters are important and their control methods are still hot in the research area.The most conventional control method of power converters is based on PID controllers combined with pulse width modulation. But this method is best suited to linear, single input, single output, unconstrained control problems. The design procedure of PID controllers for nonlinear, multiple input, multiple output, and constrained control problems is cumbersome. Besides, PID controllers can achieve satisfactory performance only in a narrow operating range; outside of this range, the performance significantly deteriorates. To overcome the inherent disadvantages of PID controllers, many new control methods are gradually proposed. Model predictive control method is one of them. The constant improvement of computational capabilities of microprocessors is enabling the wild application of model predictive control.Model predictive control method is quite flexible and inclusive. It is able to control multiple state variables, hand linear and nonlinear dynamics and easy to include constraints. MPC can predict the system states real-timely using predictive model and control through rolling optimization. It can apply to different control objects and control objectives with strong anti-interference.In this paper, the control methods and performance of model predictive control in three-phase grid-connected inverters, boost converters and two-stage grid-connected systems are introduced and analyzed. In the single zero vector current based MPC of three-phase inverter, only one zero vector is applied since the two zero vectors have the same output performance. This results in unequal loss distribution between the upper and lower switches and uneven heating. Dual zero vector voltage based MPC can balance the loss, but its control logic is complex and it is hard to program. So the dual zero vector current based MPC method is proposed in this paper. This method utilizes two zero vectors flexibly according to different switching vectors, indicating an equal distribution of switching losses among upper and lower switches and reducing switching counts with simple control logic.Based on the research of control methods for three-phase grid-connected inverters and boost converters, the model predictive control of two-stage grid-connected system is proposed. The dc-link capacitor between the two converters is used to supply stiff DC bus voltage. Because of the equivalent series resistance, the current flowing through the capacitor causes losses and heating, which may shorten the capacitor lifetime. This paper presents a detailed analysis of the capacitor current in two-stage grid-connected system and accordingly proposes an improved method with low capacitor current RMS value. The proposed improved MPC method can reduce both capacitor losses and switching losses, which contributes to the increase of whole system lifetime.Finally, model predictive control methods of three-phase grid-connected inverters, boost converters and two-stage grid-connected systems is simulated in the MATLAB/Simulink, which verified the validity and advantage of the proposed methods.