Research On Distribution Network Fault Recovery Strategy Based On Photovoltaic Output Prediction

In recent years,extreme natural disasters have occurred more and more frequently around the world,and the resulting large-scale power outages have caused extremely bad effects on human production and life,and even threatened the lives and property of the people.With the innovation and development of renewable energy technologies such as photovoltaic and wind power,its penetration rate in the power grid continues to increase.With its advantages of clean efficiency,flexible installation,etc.,the use of distributed photovoltaic and wind power in the network is an important load under extreme events Continuous power supply is an effective means to enhance the regional distribution network.The intermittent,unstable and volatile characteristics of photovoltaic output will have an impact on the power system when it is connected to the grid,which brings many challenges to grid planning,system debugging and stable operation of the power system Therefore,it is of great significance to predict the photovoltaic output.Based on the photovoltaic output prediction model,this paper mainly establishes a regional distribution network fault recovery method to conduct research.The main research content includes the following aspects:Firstly,this paper studies various meteorological factors that affect the power generation of the photovoltaic power generation system,and analyzes in detail the correlation of solar radiation intensity,ambient temperature,wind speed,relative humidity,and atmospheric pressure and other meteorological factors on photovoltaic output in a typical day.And the regularity of photovoltaic power generation under four typical weather types of sunny,cloudy,rainy and heavy rainy days.On this basis,the concept of similar days is proposed,and the gray correlation analysis method is used to extract similar days under historical samples,which improves the prediction effect of photovoltaic power generation.Then,the principal component analysis is carried out for the five meteorological factors affecting photovoltaic power generation under the same weather type.On the basis of retaining the data information,the coupling and dimensionality reduction of the data are understood,and the feature data is extracted for light The prediction model of v-output lays the data foundation.Secondly,according to the intermittent and fluctuating characteristics of photovoltaic output affected by meteorological factors,a photovoltaic output prediction model based on Hidden Markov model(HMM)theory is proposed.This paper briefly introduces the basic properties and related algorithms of hidden Markov model,uses K-means clustering algorithm to divide the historical sample data into four types of typical weather,extracts the characteristics of all kinds of weather in photovoltaic power plant through principal component analysis,obtains the main characteristic factors affecting the power as the input variables of the prediction model,and establishes the prediction model based on Baum Welch algorithm and Viterbi algorithm in HMM The model is validated by the case data of a distributed photovoltaic data in Jiaxing.Finally,based on the linear power flow model,a mixed integer linear program(MILP)model suitable for fault recovery is established.A multi-period load recovery optimization model of regional distribution network considering the constraints of different device characteristics,load status and network topology in the microgrid is proposed.Taking into account the uncertain characteristics of photovoltaic output,Gaussian mixture model(GMM)is used to model the forecast error data of photovoltaic output per hour,and the idea of quantile as risk index is proposed to establish a multi period fault recovery linear programming model with the maximum weighted load duration as the objective function.For the IEEE-34 node system where extreme faults occur and the large power grid is dissociated,the proposed multi-period load recovery strategy verifies the improvement of the resilience of the power grid in terms of the number of load restorations,load recovery power,node voltage,and the output plan of the distributed generation.