节点文献

基于二进制差分进化算法的含有分布式电源的配电网重构研究

Study on the Reconfiguration of Distribution Network with Distributed Generation Based on Binary Differential Evolution

【作者】 张冠军

【导师】 周湶;

【作者基本信息】 重庆大学 , 电气工程, 2011, 硕士

【摘要】 配电网重构是优化配电系统运行的重要手段之一,也是配电自动化研究的重要内容。目前配电网重构分为两种:一种是配电网正常运行时的网络优化重构(简称网络重构),另一种为配电网发生故障后,对非故障停电区域恢复供电的网络重构(也称为故障恢复)。本文主要对输出不受自然条件影响的微型燃气轮机和燃料电池这两类分布式电源接入配电网后的优化重构和故障恢复重构进行了重点研究。针对分布式电源并网接口的分类不同,研究了它们各自在配电网潮流计算中的模型以及处理方法。提出一种从网络末梢节点开始逐步前推到根节点的拓扑识别方法,并更新重构过程中支路改变后的配电网拓扑数据,直接形成支路的前后连接关系,并采用前推回代法来计算有分布式电源并网的配电系统的潮流。针对模拟进化类的算法容易陷入局部最优解以及收敛速度慢等问题,本文提出采用二进制差分进化算法进行配电网重构,该算法收敛速度更快,全局搜索和局部搜索能力更强。针对开关自由组合会产生大量的不可行解,导致算法效率下降的问题,本文提出了一种化整为零策略,将整个解空间划分成若干个子空间,使算法直接在解空间中进行搜索,缩减了编码维数,减少算法的不必要搜索。根据配电网重构的特点,改进了二进制差分进化算法的变异原则和交叉原则,提高算法的计算效率。以加入分布式电源后的16节点、33节点、69节点三个配电网为算例进行大量仿真,结果表明文中的重构算法是可行的。针对在含有分布式电源的配电网故障恢复重构中,需要迅速划分DG孤岛问题,本文提出一种基于分布式电源可行孤岛数据链接表的孤岛划分法,并利用二进制差分进化算法对孤岛外的非故障停电区域进行故障恢复。通过算例验证了孤岛划分方法和恢复重构算法的有效性和可行性。

【Abstract】 Reconfiguration is an important means of distribution network optimization, is also an important part of distribution automation research. Currently, distribution network reconfiguration is divided into two kinds: one is the network optimization reconstruction of the normal distribution system (referred to as network reconfiguration), and the other is seivice restoration in distribution system, for the regional of non-fault to recovery power supply (also known as Recovery) . This paper mainly studies on distribution optimizing reconstruction and fault recovery reconstruction when two types of powers which are the micro gas turbine and fuel cell access to the distribution network.According to the distributed power grid interface type this paper analyses their model in the power calculation. In the back/forward flow calculation method, taking into account the distribution network reconstruction can cause branch change, so this paper proposes a topology identification method which is from the network peripheral node began to pushing the root node, and can rapidly update the network topology data and the relationship of the link, and bring the convenient for flow calculation with back and pushing process.In view of the intelligent algorithms has shortcomings of slow convergence speed and easy converging local optimal solution, so this paper adopts differential evolution algorithm which has convergence speed faster, global search and local search ability stronger to distribution network in reconstruction.Considering the distribution network reconstruction process is the switches be opened and closed process, however, the switch’s random combination will produce large number of unfeasible solution, so this paper puts forward a method of strategy which is break whole into pieces, the whole solution space is divided into some sub-solution space, make algorithm to search solution space directly, decrease the coding dimension, reduce unnecessary search. In order to avoid producing unfeasible solution in the evolutionary process of differential evolution algorithm, this paper improved the variation operation and crossover operation to improve the efficiency of the algorithm. The results of simulation for 16 nodes, 33 nodes, 69 nodes distribution system with the distributed power show that the reconstruction algorithms is feasible.In view of recovery reconstruction of distribution network with distributed power, needs rapid dividing DG island, this paper proposes a method based on DG feasible island data link table to divided island, and use binary differential evolution algorithmthe to recover the unfault area which is outside of DG island. The results of numerical examples validate that the island dividing method and restore reconstruction algorithm are effective.

  • 【网络出版投稿人】 重庆大学
  • 【网络出版年期】2012年 01期
节点文献中: 

本文链接的文献网络图示:

本文的引文网络