Research On Probabilistic Load Flow Of Distribution Network With Photovoltaic Power Generation System

With the constant change of the global energy pattern,renewable energy has become the main energy that can be developed and utilized by human beings,especially the clean energy represented by Photovoltaic power generation has been favored by all countries in the world due to its unique advantages.In recent years,China's Photovoltaic power generation has been developed rapidly,and its installed capacity now ranks first in the world.However,there are many problems in photovoltaic power generation.For example,Photovoltaic output power has random fluctuations,and its large-scale grid connection will affect the safety and reliability of the distribution network.Therefore,in this thesis,photovoltaic power generation as the background,the photovoltaic power probabilistic model modeling method,power system probabilistic load flow algorithm including photovoltaic power supply and the impact of photovoltaic output on the distribution network voltage distribution were studied.Firstly,the traditional nonparametric kernel density estimation has poor local adaptability and boundary bias in photovoltaic power probability modeling.In this thesis,the reflection method based nonparametric kernel density estimation is combined with adaptive nonparametric kernel density estimation,an improved nonparametric kernel density estimation algorithm is proposed,and the photovoltaic power probability model is established.Through the photovoltaic power historical data,and combining the goodness of fit test and error analysis,the simulation analysis of the model shows that the photovoltaic power probability model established by this method not only effectively eliminates the boundary bias,but also improves the local adaptability and has a high accuracy.Secondly,for random variables whose distribution function is unknown or complex,the cumulant calculation efficiency is low and cannot be solved by numerical method.In the thesis,an effective probabilistic power flow algorithm is proposed by combining the improved Latin hypercube sampling with the cumulant methods.Based on the different probability distributions of the random variables,the method first uses different methods to solve the cumulant of each order of the random variables.Secondly,on the basis of the linearized AC model,the probability distributions of the system node state variables and the branch power flow are obtained by combining the Gram-charlier series expansion method.The simulation results show that the probabilistic load flow algorithm based on improved Latin hypercube sampling method not only retains the advantages of fast computation speed of the cumulant method,but also has certain accuracy and effectiveness.Finally,in order to more accurately describe the impact of photovoltaic output random characteristics on distribution network voltage distribution,this thesis adopts probabilistic load flow analysis method to carry out simulation analysis on node voltage edge distribution and joint distribution.The results show that the random characteristics of PV output have a certain impact on the voltage distribution of the distribution network,and the degree of impact is related to the grid-connected position and capacity of PV.