| With the gradual increase of the installed scale of wind power generation in China,especially the penetration rate of wind power is increasing in some areas,the contradictions and problems caused by large-scale centralized wind power generation gradually begin to appear.The decentralized wind farm near the load has become a new form of wind power integration into power grid in China.From the perspective of policy and technology,there are corresponding technical regulations and operation modes for centralized wind power and distributed wind power integration into power grid in China,while there is a lack of corresponding technical standards and related research for decentralized wind power.Therefore,this dissertation takes the decentralized wind power as the research object,and carries out the research from two aspects: the determination method of the location and capacity of the decentralized wind power access distribution network and the expansion planning of the distribution network with decentralized wind power.The main achievements of this dissertation are as follows:Firstly,the determination method of location and capacity of decentralized wind power based on the entropy weight method and grey relational analysis method is studied.The influence of decentralized wind farm connected to distribution network on voltage distribution is analyzed theoretically.Then,aiming at the shortcomings of the traditional decentralized wind power location and size determination model which does not consider the selection of candidate nodes,a candidate node selection method based on the entropy weight method and grey relational analysis method is proposed,which considers voltage stability index and loss sensitivity factor,and uses the comprehensive evaluation method combining the entropy weight method and grey relational analysis method to evaluate the candidate nodes.Then a scene reduction method based on improved k-means clustering algorithm is proposed to capture the volatility and randomness of historical statistical data and generate representative typical scenes.A multi-objective optimization model is proposed to minimize the annual comprehensive cost and the average voltage deviation.Considering various constraints,non dominated sorting genetic algorithm-Ⅱ is used to solve the optimization model,and the best compromise solution is determined based on the fuzzy set theory.It is verified by an example.Secondly,the reliability evaluation method of distribution network with decentralized wind power is studied.This dissertation introduces the reliability index system of distribution network,including component reliability parameters,load point reliability index and system reliability index.Then this dissertation introduces the component reliability models commonly used in distribution network,including two-state model and three-state model.Then,the uncertainty model of decentralized wind power output and load is established to simulate the time series of decentralized wind power output and load demand.Then,the feeder area division method of distribution network is introduced.The feeder area is used to replace the component as the minimum unit to analyze the fault impact mode of distribution network.Considering the impact of decentralized wind power connected,the fault impact classification table of distribution network with distributed wind power is established.Then a load reduction strategy based on breadth first search algorithm is proposed.Finally,a reliability evaluation method of distribution network with decentralized wind power based on Sequential Monte Carlo simulation is proposed,which lays a foundation for the expansion planning of distribution network with decentralized wind power.Finally,a distribution network expansion planning model with decentralized wind power considering equipment health index is studied.A bilevel expansion planning model of distribution network with decentralized wind power considering equipment health index is proposed.The goal of the upper level programming model is to minimize the annual comprehensive cost.The reliability level is transformed into the reliability cost,which is introduced into the planning model,and the reliability constraints are added into the objective function as penalty function.An initial grid structure generation method based on "avoiding circle method" is adopted to improve the efficiency of model solving.The lower model aims to maximize the consumption of decentralized wind power.Based on the known network structure,the operation data of distribution network are simulated,and the results are returned to the upper layer.The double nested genetic algorithm is used to solve the programming model.Based on IEEE-33 nodes distribution network,a 40 nodes system is obtained.Taking the 40 nodes system as an example,the planning results under different schemes are analyzed. |
PDF Full Text Request