Research On Load Recovery Strategy Of Distribution Network With DC Microgrid

With the progress of society,the excessive use of traditional non-renewable energy has brought severe challenges to both energy and environment.Microgrid can aggregate a variety of renewable energy to reduce the use of traditional non-renewable energy.DC microgrid has the advantages of simple structure,direct energy transformation and reliable control,which has gradually become a research hot spot in electrical field.In recent years,the frequent occurrence of power failure in the world makes people pay more attention to the safe operation and restoration of power system.The method of using DC microgrid to divide the distribution network into isolated islands as a new idea to solve the problem of distribution network fault recovery has been researched more widely.This paper introduces the structure and control mode of each part of the DC microgrid and its energy management strategy based on SOC of the battery.The DC microgrid model is built and simulated on the MATLAB/Simulink platform.An improved particle swarm optimization(PSO)algorithm was used to optimize the least square support vector machine(LS-SVM)prediction model for the load prediction in the energy management strategy.A distribution network island division method considering DC microgrid is proposed by means of simulated annealing genetic algorithm(SAGA)to solve the distribution network recovery problem.Firstly,a typical DC microgrid consisting of photovoltaic array,energy storage system and AC/DC loads is built in this paper.According to the energy requirements of different operation conditions of the DC microgrid,an energy management strategy based on SOC of the energy storage battery is designed.In this paper,the limit operating mode of the microgrid is designed based on the load/power generation prediction in a special operation condition that the microgrid is about to collapse.The specific control mode of each module under the energy management strategy is introduced in detail.Among them,the photovoltaic array is divided into two working modes: constant voltage mode and maximum power point tracking mode;The energy storage module is divided into three working modes: constant current charging mode,voltage regulation mode and standby mode.The grid-connected converter is used to control the DC microgrid to switch the working mode between grid-connected and isolated.The simulation model of DC microgrid was built on MATLAB/Simulink platform and the effectiveness of the control strategy of each part and the overall energymanagement strategy was verified.Secondly,in view of the load/power generation prediction required in the energy management strategy of the DC microgrid and its economic problems existing in the optimal operation,this paper adopts the power load data of a district in the west of China,and uses the least square support vector machine(LS-SVM)to make short-term load prediction.On the basis,particle swarm optimization(PSO)algorithm and improved PSO algorithm are used to optimize the two key parameters in LS-SVM prediction model,and the better prediction result is obtained through comparative verification.Finally,this paper takes the PG&E 69-node distribution network topology in the United States as the example,takes the situation of large area power loss due to the fault of the distribution network as the background,takes the recovery of the most important load and the disconnection of the least switches as the main objective,fully considers the impact of load priority and controllability,and uses the SAGA algorithm to make the distribution network island division plan with DC microgrid,so that the island network can recover important load as much as possible and minimize the number of disconnected switches at the same time.MATLAB platform is used for programming to verify the scheme,and the superiority of this method are verified by comparing with other schemes in the paper.