Design Of Three-Phase Grid-Connected Inverter Control System Based On BP Neural Network

Photovoltaic grid system is an important part of using solar energy and converting to electrical energy.In this system,the grid-connected photovoltaic inverter produces a large number of harmonic components during the converter process,and the switching of different types of load and the fluctuation of the power grid will have a very significant impact on it.If not controlled,it will cause the whole system instability.The ability of inverter to convert and transfer electric energy,and the effect of following the grid voltage have become important factors to determine the stability and requirements of photovoltaic grid-connected system.The control of grid-connected inverter has become one of the research focuses in the new energy field.Current control is the main control mode of grid-connected inverter system.It requires more sensors to sample various parameters,which increases the equipment volume and cost,and adds more uncertainties and danger to the system.In grid-connected inverter system,the current controller generally adopts PI controller.The controller parameters need to be obtained through experience testing,which brings inconvenience to the realization of the control system.On the basis of above problems and current studies,this study takes the LCL type three-phase grid-connected inverter as the research object,and makes an in-depth study on grid-connected current control structure and current controller.Firstly,the circuit structure and mathematical model of grid-connected inverter,modulation method of switch tube of three-phase inverter,characteristics of different types of filters,active damping method and passive damping method,current controller are analyzed.The structure and controller type of grid-connected inverter control system are also selected.Secondly,traditional current control structure is analyzed and improved.Based on the active damping method,the control structure of capacitor branch current feedforward and inverter output current closed-loop is established.This structure protects the inverter switch from overcurrent,suppress the harmonic component of grid-connected current,and improve the grid-connected current quality and system stability.Thirdly,the PI current controller under BP neural network is designed to set the controller parameters online to improve the control effect.Finally,a Simulink simulation model is built on the MATLAB experimental platform.This control system is simulated to verify the overall harmonic distortion rate of grid-connected current and the effect of grid voltage following.