| The structures distribution networks are closed loops structure in the most cities, but when they are operating the loops will be opened. In this way, the distribution network reconfiguration can be realized by opening or closing the section breakers or tie breakers in order to reduce network losses, insulate faults, displace loads etcWith the establishing of Chinese power market, it has become one of the important problems for the city power companies in our country to enhance the economy benefit and improve the quality of power supply, and that distribution network reconfiguration is the important means to improve the economy benefit and security of distribution power system. At present, the developments of distribution power system automatization and load forecast techniques have made distribution network reconfiguration become possible.Up to now, inside and outside the country, among all of the optimization algorithms that have been proposed for solving the distribution network reconfiguration problem, the OFP, branches exchange algorithm, ANN, ACO, SA, GA, Tabu and expert system are commonly used. In these algorithms because GA has preferable capability and can get the best solution of overall situation, it has been drawing more and more experts' attention.Now when GA is used to solve the distribution network reconfiguration problem, the main problems that it is faced are how to enhance the calculation efficiency, eliminate the "infeasible solutions", and improve its practicability. For solving these problems, the author read a lot of correlation literatures, analyzied the algorithms that have been proposed and the application requirements of distribution network reconfiguration. Based on these, the author builds the maths model for distribution network reconfiguration, proposes the notions of "special branch", "special node" etc and their storing means; analyzes the harm of "infeasible solutions" and compares the methods have been proposed for solving this problem; proposes the genetic algorithm based on "improvement" strategy for power distribution network reconfiguration and its realization technique, which can solve the "infeasible solutions" problem effectively and has relatively high search efficiency; for enhancing the practicability of the algorithm, the author also proposes the dynamic encoding and decoding technique and their realization methods, which can be used with fore back super session power flow algorithm. At last, a program based on thesealgorithms and techniques is presented, in which the solutions reserve and grouping calculation strategies are used.At the end of the paper the solutions of three standard test computations are shown, in which network losing is minimized by distribution network reconfiguration. They make clear that the algorithms and techniques proposed in this paper can be effectively used in solving the distribution network reconfiguration problem; the structure of the program is reasonable and the writing of program is right.The genetic algorithm based on "improvement" strategy for power distribution network reconfiguration proposed in this paper not only can be used in distribution network reconfiguration, but also can be used in the power source selection optimization in power company business extension, the design of distribution network, and the improvement of voltage level. It has vast range of application prospects. The dynamic encoding and decoding, the solution reserve and grouping calculation strategies lay a foundation of its practical application. |
