配电网络重构的方法研究

发布时间：2018-01-19 11:32

本文关键词： 配电网络重构潮流计算量子粒子群算法仿射传播聚类算法　出处：《山东大学》2017年硕士论文　论文类型：学位论文

【摘要】：在不断推进智能电网发展的大背景下,建设安全、可靠、经济运行的配电网络逐渐成为人们所关注的焦点。由于配电网络中各节点的电压等级比较低,导致它的网损成为电力系统网损中不可忽视的一部分。配电网络重构作为一项重要的配电网自动化技术,可以在不投入更多额外设备的情况下,仅对线路上开关的状态进行调整,就可以实现降低网络损耗,改善电压质量的目的。在实际配电网络运行过程中,负荷状态是实时变化的,这是配电网络重构中一个不易处理的问题。本文先从单个时间断面的角度,利用随机类算法对配电网络中负荷状态保持不变的情况进行分析。然后再从整个时间区间的角度,根据负荷状态实时变化的情况,提出一种配电网络动态重构方法。首先,本文对配电网络重构问题中的基本理论作了详细介绍,并给出了相应的潮流计算方法和数学模型。通过引入节点分层矩阵Layer M以及相应的父节点矩阵NU来描述网络层次关系,结合前推回代法,实现同一层中数值的并行计算,从而提高了计算效率。然后把降低网损作为配电网络重构的最终优化目标,并给出了相应的目标函数和约束条件。其次,本文提出了一种基于IQPSO算法的配电网络静态重构方法。该方法二考虑到QPSO算法的收敛性会受到初始种群分布情况的影响,所以引入了Logistic混沌模型来提高初始种群的遍历能力;另外,通过对收缩—展系数α进行改进,提高了 QPSO算法的动态自适应性。在配电网络静态重构的过程中,根据网络中的回路情况,采用十进制的编码策略对粒子向量进行编码,降低了重构过程中不可行解产生的比例;通过对网络中节点的层次关系进行分析,给出了重构解可行性的判断方法以及相应的处理方式;最后,介绍了该静态重构方法的具体步骤。通过仿真计算,证明了该静态重构方法具有正确性和快速性。最后,本文提出了基于仿射传播聚类算法的配电网络动态重构方法。该方法考虑到配电网络节点负荷变化的特点,先将所研究的时间区间等间隔地划分为若干个子区间,并认为各子区间中的负荷情况保持恒定不变,然后利用仿射传播聚类算法对整个时间区间内的负荷状态进行聚类,得到聚类代表点以及相应的聚类情况,并根据聚类代表点的负荷状态进行若干次静态重构。同时,对聚类结果中那些孤立的时间断面作修正处理,以实现减少开关操作次数的目的。该方法通过节点负荷状态的内在特征对整个时间区间进行划分,可以有效避免主观因素对划分结果造成影响。通过仿真计算以及对计算结果进行分析,验证了该动态重构方法的正确性和合理性。
[Abstract]:Under the background of continuously advancing the development of smart grid, the construction of safe, reliable and economical distribution network has gradually become the focus of attention, because of the low voltage level of each node in the distribution network. Distribution network reconfiguration, as an important distribution automation technology, can not be ignored without additional equipment. The purpose of reducing network loss and improving voltage quality can be achieved by only adjusting the state of switch on the line. During the operation of actual distribution network, the load state changes in real time. This is a difficult problem to deal with in distribution network reconfiguration. The random class algorithm is used to analyze the condition that the load state in the distribution network remains invariant, and then from the angle of the whole time interval, according to the condition of the real-time change of the load state. A dynamic reconfiguration method for distribution network is proposed. Firstly, the basic theory of distribution network reconfiguration is introduced in detail. The corresponding power flow calculation method and mathematical model are given. By introducing the node hierarchical matrix Layer M and the corresponding parent node matrix NU to describe the hierarchical relationship of the network, combined with the forward pushback method. The parallel computation of numerical value in the same layer is realized, and the efficiency of calculation is improved. Then, reducing network loss is regarded as the final optimization goal of distribution network reconfiguration, and the corresponding objective function and constraint conditions are given. Secondly. This paper presents a static reconfiguration method for distribution network based on IQPSO algorithm. The second method considers that the convergence of QPSO algorithm will be affected by the initial population distribution. So the Logistic chaotic model is introduced to improve the traversal ability of the initial population. In addition, the dynamic self-adaptability of QPSO algorithm is improved by improving the contraction-expansion coefficient 伪. In the process of static reconfiguration of distribution network, according to the loop situation in the network. The decimal coding strategy is used to encode the particle vector, which reduces the proportion of the infeasible solution in the reconstruction process. Through the analysis of the hierarchical relationship of nodes in the network, the method of judging the feasibility of the reconstruction solution and the corresponding processing methods are given. Finally, the concrete steps of the static reconstruction method are introduced. The simulation results show that the static reconstruction method is correct and fast. Finally. This paper presents a dynamic reconfiguration method for distribution network based on affine propagation clustering algorithm, which takes into account the characteristics of load variation of distribution network nodes. First, the time interval is divided into several sub-intervals at equal intervals, and it is considered that the load in each sub-interval remains constant. Then, the load state in the whole time interval is clustered by affine propagation clustering algorithm, and the representative points of clustering and the corresponding clustering situation are obtained. According to the load state of the cluster representative points, the static reconstruction is carried out several times. At the same time, the isolated time sections in the clustering results are corrected. In order to reduce the number of switching operations, the method divides the whole time interval by the inherent characteristics of the node load state. It can effectively avoid the influence of subjective factors on the partition results. The correctness and rationality of the dynamic reconstruction method are verified by simulation and analysis of the calculated results.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM76;TP311.13

【相似文献】