Operation Risk Analysis And Topological Optimization Of Distribution Network Considering The State Of Distribution Terminals

With the gradual penetration of uncertain factors such as flexible load and distributed power supply in distribution network,the operation state of power grid also flows into random characteristics.How to comprehensively evaluate the possibility and severity of the abnormal state of the power grid and realize the risk assessment of power grid operation is of great significance to ensure the reliable and high-quality power supply of the distribution network and the optimal operation of the power grid.However,the existing distribution network operation risk assessment methods generally have some problems,such as incomplete consideration of uncertain factors,complicated calculation of risk analysis methods or large errors.The power supply path optimization based on risk analysis also has some problems,such as too large search space and poor real-time performance under large node network.Based on the above problems,this paper carried out the following work:(1)Firstly,the mapping method between uncertain characteristics and risk categories of distribution network is studied by using complex valued CNN(cv-cnn).Using the characteristic function theory to realize the matrix modeling of complex uncertain random variables,the complex input of cv-cnn is constructed,the risk of voltage and power exceeding limit is calculated by the time sequence discrete sampling of daily power curve of load and distributed power combined with Monte Carlo simulation(MCS),the risk sample space is constructed,the risk category label is formed by clustering,cv-cnn is trained,and constructed in front of the output layer of cv-cnn Feature conversion layer to ensure the conversion of complex features of hidden layer to real field risk categories.(2)Secondly,the reliability model of bulk distribution terminals and the state evaluation method of individual distribution terminals are studied.For the same batch of terminal samples,the fault samples are expanded by BP neural network and empirical reliability function,and then the regression of reliability function parameters is realized.For individual distribution terminals,the algorithm of Apriori data mining combined with fuzzy comprehensive evaluation is proposed to realize the rapid evaluation of equipment status of individual distribution terminals.Through Apriori algorithm,the association rules of each function module are mined from multi-dimensional state parameters,and the corresponding state evaluation indexes are established.The comprehensive objective weight of each index is determined by the credibility and promotion degree of association rules,and the fast and effective quantification of the state of terminal function layer and device layer is completed in turn by combining with fuzzy comprehensive evaluation.(3)Thirdly,the risk analysis and evaluation method of power grid integrated operation is studied.This paper analyzes the influence of terminal equipment abnormal state on the expansion of fault partition and isolation scope,establishes a distribution terminal spatial state distribution model to represent the probability of abnormal operation state and the spatial distribution of terminal equipment,and independently quantifies the distribution network operation risk analysis results under the single disturbance of terminal equipment abnormal state.The model of comprehensive risk assessment is constructed to realize the numerical assessment of comprehensive operation risk including the influence of distribution terminal state.Through the transfer learning,the transfer from the parameters of the classification model to the regression model is realized,and the training process of the whole model is accelerated.(4)At last,a fast searching method for the optimal topological structure of power grid is studied to minimize the comprehensive operation risk.Based on the results of comprehensive risk assessment,the distribution network topology optimization model aiming at minimizing the short-term and daily cumulative comprehensive operation risk of distribution network is analyzed and established respectively,and the topological feature constraint of distribution network is introduced to reduce the redundancy of feasible solution space,so as to ensure that the searched topological open-loop combined solution meets the effective conditions of open-loop operation and no island formation of distribution network,which greatly reduces the The time consumption of the invalid switch combination solution,combined with cultural gene algorithm,completes the fast and accurate search of the optimal switch combination strategy to determine the topological structure.