Research On High-efficiency Non-isolated Photovoltaic Grid-connected Inverter System

In virtue of the serious energy crisis and environmental pollution problems, the energy of human choices tends to be clean and renewable energy. Solar energy is the most clean and one of the most widely distributedly renewable energy. Therefore, it has been widely concerned by governments and scientific researcher, become the main development tide of solar energy utilization, and brought big economic benefits to human society. In addition, the emergence of high-frequency, low-power switches and high-performance processor allow more advanced control strategies to be used in photovoltaic grid-connected inverter system which improve inverter system conversion efficiency, reduce the cost.weight and volume, prolong the life of the service and increase the reliability of the system. So photovoltaic grid-connected inverter system can be used in large scale, In this paper the core control algorithm of solar energy photovoltaic power generation system is studied in order to reduce the THD of grid-connected current and improve the conversion efficiency of system. By using H4 full bridge, this paper carried out theoretical study and simulation verification for the grid voltage feed-forward PI control strategy and model predictive control strategies of the solar grid-connected power generation system.This paper firstly introduces the grid voltage feed-forward PI control strategy and some traditional model predictive control strategy, describes the relevant theory in detail, and carries out simulation verification and comparative analysis. Grid voltage feed-forward PI control, direct digital model predictive control, improved model predictive control and model predictive control strategy with unconstrained optimal pulse width modulation are based on the traditional SPWM modulation control method, and their control performances are affected by PWM modulation.In previous work, there are current static error and poor dynamic performance in the voltage feed-forward PI control strategy, and the poor current static error in the traditional model predictive control strategy. This paper prosposes a multistep model predictive control strategy based on sphere decoding for single-phase inverter, focuses on introducing the sphere decoding mechanism and the binary tree traversal method, and carries out the simulation. Simulation results show that compared to the grid voltage feed-forward PI control and traditional model predictive control strategy, multistep model predictive control strategy based on sphere decoding in the single-phase photovoltaic grid-connected inverter application decrease searching steops largely compared with enumeration method, and obtain lower grid-current steady-state error and improve anti-interference ability.