Research On Distribution Network Reconfiguration Based On Refined Genetic Algorithm

Distribution network reconfiguration is not only an important method for circulating and controlling of distribution system, but also an important part of distribution management system. In theory, there exists a best network configuration for any distribution network. Under the best network configuration, the voltage at each load point, power losses and load balance are all better than other feasible methods. Generally, distribution networks are built as interconnected meshed networks, while in the operation they are arranged into a radial configuration. There are a number of normally-closed switches along the feeders and a number of normally-opened switches between the feeders. Modify the network configuration by changing the open/closed status of the sectionalizing and tie switches under both normal and abnormal operating conditions is to improve the security, economy and reliability of the distribution system.In theory, the distribution network reconfiguration is a complex, multi-objective, non-linear, combinatorial optimization problem. From 1980s, the distribution network reconfiguration has been extensively studied and many relatively mature methods and theories of network configuration has formed and developed, but all methods have shortcomings on their ability of global searching or speed of convergence. The ability of global searching and extensive application in other scopes of genetic algorithm has proved that it is an attracting goal for us to study how to apply genetic algorithm (GA) to distribution network reconfiguration.First, the method of network topology analysis is discussed. On the basis of analyzing the current techniques for power flow calculations of distribution network, according to the feature of distribution systems, the forward/backward sweep method based on the branch power is used to calculate the power flow of distribution network. The model of the distribution network reconfiguration is researched and discussed. Taking load balancing into account, the mathematics model aimed at minimizing line losses is adopted in the thesis.The characteristics of the genetic algorithm is analyzing in detail. Now there are many problems when genetic algorithm is used to solve the problem of distribution network reconfiguration. For solving these problems, some improvements are put forward as follows. In order to avoid premature convergence and improve the speed of convergence, an adaptive genetic algorithm whose evolution is based on the stage is put forward. This thesis lays emphasis on the improvements of population size, selection operator, crossover operator, mutation operator, termination conditions and etc, at the same time adds into inversion operator. To solve the problem of infeasible solutions in distribution network reconfiguration, the methods of encoding, building initial population, crossover, mutation and inversion are improved, and through these improvements the infeasible solutions reduced evidently. A program of the proposed refined genetic algorithm (RGA) is presented. The proposed method is applied to a typical tested system and a practical distribution system, and the results have shown that the ability of global searching and the speed of convergence of the RGA are remarkably improved, and the RGA can be effectively used in solving the distribution network reconfiguration problem.The load of the practical distribution system is varying dynamically along with time, which makes the optimal configuration change at any time. Dynamic distribution network reconfiguration is required to ensure safety, quality and economic of the distribution system during the load varying in operation periods. The developments of distribution system automation and load forecast techniques have made dynamic distribution network reconfiguration possible. At last, the model, the characteristics and the solution of dynamic distribution network reconfiguration were discussed.