Distributed DC Microgrid Based On Consensus Algorithm Control Research

In recent years,clean energy has received more and more attention,and the interconnection control between microgrids has also become a research hotspot for a large number of scholars.Therefore,the research on the interconnection control of microgrids based on clean energy is of great significance.The DC microgrid can connect energy storage units such as photovoltaics and batteries through a distributed control method,and connect it to the power system through a power electronic converter,thereby providing load demand for users.The proportional distribution of power among the distributed energy sources and the maintenance of bus voltage stability are the two main control objectives of the DC microgrid.In order to achieve this control goal,this paper connects the energy storage unit and the DC bus through a converter,discusses the control issues,and conducts model verification and indepth research according to different scenarios.First of all,in order to ensure the quality of power transmission,the energy storage element needs to be connected to the bus terminal through the converter.By studying the energy storage unit and the converter topology,design and determine that the primary control layer of the microgrid adopts voltage outer loop current inner loop double closed-loop control method,and then build simulation model of each converter unit to verify the effectiveness of its control method.Then,on the basis of determining the primary control method,to ensure the coordinated operation of the interconnected DC microgrid,it is necessary to overcome the power distribution problems and voltage deviation problems of each branch caused by uneven line impedance.Therefore,this article improves on the traditional droop control and designs an adaptive droop control method,which can adjust the droop coefficient by itself according to the operating current of different loads.The algorithm is based on the theoretical basis of the consensus algorithm,through the collection of local information and neighbor information,so as to provide correction for the droop coefficient,and synchronize with the local current.The algorithm can improve the accuracy of load ratio sharing and further adjust the bus.It also solves the power distribution problem caused by uneven line impedance that cannot be solved by the traditional fixed droop coefficient.Based on the adaptive droop algorithm,in order to compensate for the natural voltage drop caused by droop control and maintain global voltage stability,a voltage correction amount is added in the secondary control link.First,obtain the neighbor node voltage information through the voltage observer,obtain the estimated average value of the bus voltage,and generate the voltage correction value,which adjusts the reference voltage of the droop control mechanism to generate a new voltage reference value.In the overall secondary-level control algorithm,not only different line impedance issues are considered,but high-density communication connections are not required.In order to determine the effect of the control algorithm on bus voltage stability adjustment and load ratio sharing,four groups of interconnections are built in Simulink.The simulation model is verified for different scenarios.The paper verifies the plug-and-play of the DC microgrid and the dynamic performance when a single link fails under the effect of the adaptive droop control method and the bus voltage recovery control measurement.Finally,considering that the development of DC microgrids in a large-scale direction will cause a rapid increase in the amount of data,and there are also problems that it is difficult to model complex power electronic devices,so a data modeling method based on extreme learning machines is designed.This method can not only solve the problem of large-scale data volume modeling,but also predict the operation status of the DC microgrid at the next moment.In order to prove the validity of the data model,the operating status of the microgrid was predicted by collecting data,and modeling and analysis were performed on different scenarios.Through the analysis of the model training process and the calculation of the fitting accuracy,the validity of the data model is proved.