Distribution Network Topology Pattern Analysis And Automation Model Research

Distribution network which possess the fetures of diverse topological structure, flexible operation mode is located in the end of the power system. It fulfill the responsibility of power distribution, and affect the user power supply reliability and power quality directly, and implement flexible and interactive smart power, the diverse topological structure and operation and flexible features. According to the nearly five years running statistics come from Guangdong Power Grid Corporation, more than 90 percent of the power outage and more than 80% of the user annual outage time due to the distribution network, and the ratio of distribution network fault exceed 35%. As the core of the distribution automation, feeder automation technology which can effectively reduce troubleshooting time, is increasingly taked seriously by Grid Corporation, and is steadily advanced of large-scale application. Currently, distribution automation planning mainly rely on the subjective experience of planners, and lack of decision support tools. It is difficult to achieve the overall optimum of economy and reliability. Therefore, this dissertation aim at several issues of distribution automation for scientific exploration. The main research work and results are as follows:1. To solve the problem of carrying out fine analysis of large-scale distribution network distribution automation planning, the feeder is characterized as multi-way tree containing node weights according to its constituent elements, and establish a heavy feeder tree model. And more, this dissertation propose a distribution network topology similarity algorithm based on the tree edit distance. The algorithm which is based on the tree edit distance algorithm of unordered rooted labeled trees, make use of the cost of change, insert and delete to obtain the absolute distance between the topology editor, and then to solve similar metrics normalized. Numerical results show that the proposed algorithm can effectively get topology similarity, and the calculate time complexity of the similarity metrics is the same as tree edit distance.2. To plan guidelines for the development of automation and improve the efficiency of automated planning, this dissertation present the topology spectral clustering algorithm based on the similarity matrix, and then abstract the applicability of automation solutions. The algorithm is capable of distribution network topology collection grouped into clusters composed by a series similar topologys; and then propose an integrated cluster similarity index to position typical feeder template, as the cluster’s representative. To rationalize the effection of clustering, this dissertation solve the multi-tree node weights by differential evolution algorithm, in order to achieve further rapid collation and non-sample feeder automation solutions identified. Numerical results show that: the mentioned clustering algorithm can find the typical topology quickly.3. To reconcile the conflict of feeder automation transformation effects and investment costs, this dissertation present a multi-objective mixed integer programming model, whose objective function is systematic annual outage time and investment costs, and constraints consist of switch settings, the number of path breakers, RMU configuration, and troubleshooting outage time. The algorithm is based on standardized and efficient method to obtain accurate and uniform distribution of the pareto frontiera completely, and then determine the optimal solution according to user preferences compromise decision function. Numerical results show that: the proposed model is suitable for automated planning decisions quickly; and headend installed in the branch line circuit breaker can be a good user of the annual outage time shortened, and in the trunk line disconnect switches installed "three remote" automatically switch can further short the user’s average annual outage time.4. For the problem of the scarcity of distribution automation planning decision support tools, this dissertation develop the distribution automation planning decision support software based on Auto CAD/Object ARX. The software includes many functional modules, such as electrical equipment/components of automatic identification, parameter settings, network topology identification, distribution network reliability assessment, and distribution automation planning decision. The software has the characteristics of clear interface, simple operation, and easy-to-use, and it can be run on x86/x64 for Windows XP/7/Vista/8 platform. Combined with the actual situation in Shenzhen Pingshan distribution network, the software achieve the application and verification. Tests show that: the software lays the foundation for the automation of planning to carry out the work.This dissertation is supported by National Natural Science Foundation of China(51307063); National Key Technology Research and Development Program of the Ministry of Science and Technology of China(2013BAA01B02).