【作者】 韦肖燕；

【导师】 周羽生；

【作者基本信息】 长沙理工大学 ， 电力系统及其自动化， 2009， 硕士

【摘要】 随着遗传算法、蚁群算法、神经网络算法等人工智能技术的引入,配电网故障判断领域的研究进入了一个新的阶段,由于遗传算法和蚁群算法各自存在的缺陷快速准确地实现故障区段的判断始终是一件比较困难的事情。为此,本文尝试将遗传算法和蚂蚁算法融合的GAAA算法引入到配电网故障定位算法的研究中。首先利用遗传算法的随机搜索、快速行、全局收敛产生故障区段的初始信息分布。然后,充分利用蚂蚁算法的并行性、正反馈机制以及求解效率高等特性,根据遗传算法迭代后的最优组合形成馈线初始故障信息素分布,将遗传算法求解结果转换的信息素值代入蚁群算法的求解步骤中。本文利用树状24馈线的配电网络发生2处故障时的情况对算法进行了仿真,通过对仿真所得数据进行分析比较,证明了GAAA算法的有效性。本文提出了多电源多点故障的情况下故障网的分离方法,同时还提出了馈线网的分离,避免了误判情况的出现。同时对评价函数也进行了改进,使得多电源多点故障下的配电网故障判成为可能,在本文中还探讨并实践了用遗传算法结果产生的选择概率来转换蚂蚁算法信息素值的计算方法。较传统的人工智能算法相比较,本文GAAA算法的引入有效的提高了配电网故障定位的定位速度和定位效率,使得算法能够应用在多电源多点故障的网络中。更多还原

【Abstract】 With the used of genetic algorithm, ant colony algorithm and neural network algorithms as well as the artificial intelligence technology, the research of the fault judgment of distribution network entered a new stage. Because of the limitation of the genetic algorithms and ant colony algorithm, it is still a difficult thing to determine the fault area fleetly and correctly. In order to solve the problem, we try to use the GAAA algorithm to determine the fault section of the distribution network, which integrates the advantage of the genetic algorithms and ant algorithm.First , we use random search, fast line, and overall convergence to obtain the initial distribution massage of the fault sections, which is a genetic algorithm process. Then according to the iterative Optimal portfolio of the initial fault massage of the feeder, the massage value conversed by the genetic algorithms is then used to the solving of the problem in the ant algorithm process, by the advantage of its parallel and positive feedback mechanisms, as well as high efficiency.In this paper,a tree of the distribution network of 24 feeders in which two failures are occurred is used to simulate the algorithm, the simulation dates are compared and analysis, and the effectiveness of the GAAA algorithm is proved. We present a method of failure determination of the distribution network under the condition of multi-power multi-point in this paper, and we also present a method of feeder network separation, to avoid the emergence of a misjudgment of the situation. At the same time, the evaluation function is also improved too, which makes multi-power multi-point distribution network fault-contracting possible. The method of the massage value of ant colony algorithm conversion is also discussed and practiced, using the selection probability produced by the genetic algorithm. Compared to the traditional artificial intelligence algorithm, GAAA algorithm improve the speed and efficiency of fault location of the distribution network, and it can be used in multi-power multi-point failure of the power network.更多还原