【作者】 卫志农；

【导师】 鞠平；

【作者基本信息】 河海大学 ， 电力系统及其自动化， 2004， 博士

【摘要】 随着电力市场的逐步形成，电力用户对电力供应的要求越来越高，为了向用户提供供电可靠、质量优异、价格合理的电能，实现配电自动化是最重要的手段。为了充分发挥配电自动化系统的作用，提高配电系统运行的安全性和经济性，并能够进行正确的分析与决策，首先必须正确全面地掌握配电系统的运行状态，并向其它应用软件提供可靠而全面的实时运行方式。本文提出了配电系统运行状态分析的概念，并运用现代数学分析方法和人工智能理论，重点对配电系统运行状态分析进行研究。 本文的主要研究工作如下： 1．在研究现有配电系统状态估计的基础上，根据配电系统的特点，提出了基于相坐标系的配电网三相不对称快速解耦状态估计算法。该算法是在等效量测变换的基础上，引入输电网三相不对称潮流计算中的解耦—补偿电流模型，利用状态估计迭代方程的特性，可以实现三相不对称状态估计方程在相坐标下完全解耦，加入旋转变换，可进一步实现法方程的实、虚部严格解耦。该算法既保留了等效电流变换法的优点，同时又改进了其不足，具有良好的应用前景。 2．不良数据检测与辨识是配电系统运行状态分析的主要功能之一。本文针对配电系统中量测数据的不确定性，运用现代模糊数学理论，提出了一种新的辨识不良数据的传递闭包方法。它采用标准残差r_N和相邻采样时刻量测值之差△Z作为两个特征值进行模糊聚类分析，用标定公式计算出量测值属于不良数据的相似系数，进而用平方法得到模糊等价矩阵，再根据λ截矩阵准确辨识出不良数据。算例证明，该方法有效可行。 3．在研究现有配电网络故障定位算法的基础上，本文提出了两种配电网故障定位的算法：配电网故障定位的改进统一矩阵算法和配电网故障定位的高级遗传算法。配电网故障定位的统一矩阵算法，针对传统统一矩阵算法的不足，提出采用有向图描述配电网络，从而避免了矩阵相乘和规格化，大大改进了现有算法。高级遗传算法通过构造新的评价函数，建立了一种新的更适合于配电网的数学模型，不仅可以避免误判，准确定位，而且具有更强的容错性能。摘要 4.在研究现有各种配电网络重构算法的基础上，本文提出了基于动态粒子群优化算法的配电网动态重构。它以配电网络电压稳定为目标函数，采用可以动态跟踪环境变化的改进粒子群算法对现有典型的辐射状网络进行动态实时重构，并提出了一种新的编码形式，按照系统的拓扑结构划分粒子种类，从而利用改进粒子群算法的社区特性避免过早收敛。经过与遗传算法进行比较，结果表明该方法步骤简单，操作方便，适合于对计算时间要求更高的工程应用。关键词:配电系统、运行状态分析、状态估计、不良数据检测与辨识、故障区间定位、动态重构、模糊理论、动态粒子群优化算法更多还原

【Abstract】 With the gradual formation of the electricity market, the customers’ demands for supply of power have become much greater than ever before. The most important way to supply the reliable, high-grade, appropriate-priced electric energy is to employ the distribution system automation (DSA). To display sufficiently the functions of DSA, improve security and economical efficiency of the running of distribution system, and achieve a sound analysis as well as an accurate decision-making process, at first, it’s necessary to master the running state of distribution system and supply the reliable, overall and real-time running module to the other application software. In this paper, a concept of running state analysis of distribution system is defined, and the modern mathematic analytical methods and artificial intelligent theories are used to study especially the running state of distribution system. The chief research work of this paper is as follows:1. On the basis of state estimation of distribution system in possession, according to the characters of distribution system, a new algorithm of the fast decoupling asymmetrical three-phased state estimation for distribution system on phase coordinates is proposed and the decoupling compensation modeling which is used for asymmetrical three-phased load flow is introduced in this paper. According to the characters of iterate equation, the equation of asymmetrical three-phased state estimation can be decoupled strictly on phase coordinates, and with the rotation transformation added, the real and image part of normal equation can also be strictly decoupled. The algorithm will not only save the advantages of equivalent current measurement varying algorithm, but also improve its disadvantages, so it has better application in the future.2. Bad data detection and identification is one of the chief functions in the running state analysis of the distribution system. Because of the indeterminacy of measure data in distribution system, a new method for identifying bad data-transitive closure method is proposed by using the modern fuzzy theory. It uses the standardresidual rN and the difference △z of measure values between adjacent sampling houras two kinds of eigenvalues to do fuzzy and clustering analysis. With the calibration formulation, the similarity coefficients which describe the extent of measure values belonging to bad data can be calculated, and the fuzzy equivalent matrices can be obtained via square method, then the bad data can be identified via λ cut matrices. Theexample proves this method effective and feasible.3. On the basis of fault location in distribution system in possession, two kinds of methods of fault location in distribution system are proposed in this paper: a unified matrix algorithm for fault location and a refined genetic algorithm for the fault location in distribution system. Because of the disadvantages of traditional unified matrix algorithm, the first one uses the directed graph to describe distribution system, and in this way it can avoid matrices multiplication and normalization. The second one creates a new evaluation function, and set up a new and more suitable mathematic modeling for distribution system, it not only avoids misjudge, but also has more powerful fault-tolerance.4. On the basis of reconfiguration algorithms of distribution system in possession, a new reconfiguration algorithm based on dynamic particle swarm optimization (DPSO) is proposed in this paper. It adopts the voltage stability of distribution system as the object function, and uses the improved particle swarm optimization algorithm which can trace dynamically the environment’s change to make the real-time dynamically reconfiguration of distribution system, and proposes a new coded form which divides the particle module according to the topological structure of distribution system, and then avoids the convergence precociously via the community characters of DPSO. Finally, it is compared with the genetic algorithm, the result of which sh更多还原