The Optimal Planning Of Power Distribution Network Based On Connection Mode

Distribution network planning is concerned with the optimal selection of the building time, building location and the wire diameter of the network; the obtained network should reach the best results on the aspects of economy, reliability and so on; and the obtained network should be subjected to the constraints such as feeder capacity, voltage drop and power balance and so on. It is a complex NP-hard optimal problem with characteristics such as multi-objects, multi-stages, non-linear and so on. Especially if the distribution network connection modes and even the transition of connection modes should be taken into account, the condition will become more complex and the difficulty will be increased greatly. Intelligent optimization algorithm, multi-objective and multi-stage optimal theory are adopted in this paper to study the problems deeply as follows.1) A bidirectional searching method for reliability evaluation of complex distribution network is proposed in chapter 2, to prepare for the reliability calculation in the chapters following the chapter. The topological relationship of network is represented by tree structure, and the acquiring of distribution network reliability indices is divided into two steps: first, the bidirectional searching is executed, and various measures are taken to get several line musters with different attributes; then, the line musters are processed, and the results are used to calculate the reliability indices. With the new method, the reliability indices of a complex distribution network with sub-feeders and alternative suppliers can be calculated conveniently, and the changing of switch installation can be treated flexibly. And it is proved to be feasible and valid through examples.2) A partheno-genetic algorithm based on tree structure encoding for (multi-level branch) radial distribution network optimal planning is presented in chapter 23.2, and two new genetic operators such as shift operator and redistribution operator for tree structure encoding partheno-genetic algorithm are proposed in this chapter, after the reason that the convergence of bintree structure encoding genetic algorithm can be generalized to tree structure encoding genetic algorithm has been given. Prim algorithm is employed to produce preliminary radial networks, which are better than entirely random schemes. The advantages of partheno-genetic algorithm are fully utilized in distribution networks optimal planning. All schemes in the solving process are always naturally being radial pattern, no need for inspection of connectivity and being radial in connection of the networks. The determination of wire diameter, the treatment of street cross points and the method of expansion planning for distribution networks is discussed. The examples of distribution network planning show that the method is feasible and efficient. Finally, the mutation operator has been proposed to perfect the partheno-genetic algorithm based on tree structure encoding.3) The tree structure encoding partheno-genetic algorithm is combined with estimation of distribution algorithm to get a hybrid algorithm to solve the problem of distribution network planning. In the hybrid algorithm, the search direction of partheno-genetic algorithm (PGA) can be estimated by estimation of distribution algorithm (EDA) and there is no immature convergence in PGA. By the complementary advantages of the hybrid algorithm the convergence rate is accelerated and the hybrid algorithm is more steady and effective.4) The ideas of connection mode space, connection mode subspace, templet space and so on, have been proposed in chapter 4. The PGA has been used to optimize the planning of power distribution network based on connection mode space (except the multi-level branch radial distribution network). The chromosome which stands for the distribution network is expressed by the object of compressed matrix class. The configuration-unit and its relative information are used to initialize the distribution network under a given connection mode. The exchange operator, shift operator, configuration-unit increase and decrease operator, redistribution operator and so on have been designed to execute genetic operation.5) The multi-objective optimization theory has been adopted to compare and analyze the economy and reliability of some common connection modes of 10kV MV distribution network. The synthetical fuzzy evaluation indices for distribution network have been given in chapter 35.5. The method of deciding weight factors of indices, and the method of deciding fuzzy membership function and relative parameters have been discussed in the chapter. Some development parks with distinct characteristics have been investigated, and the load compositions of these typical development parks have been analyzed. Some common 10kV MV distribution network connection modes are considered as study objects, and the optimal distribution networks have been obtained by the multi-objective PGA. Then the best connection modes for the typical development parks have been obtained, by the comparison of all the evaluation indices.6) Based on the previous several chapters, the multi-stage optimal planning model of distribution network has been established in chapter 6. In the planning model, the transition of distribution network connection modes under uncertainty condition of load has been taken into account. A special treatment method for multi-stage distribution network planning has been proposed, by which the transition between two kinds of connection modes in different planning stages can be carried out easily.