【作者】 邬林勇；

【导师】 钱清泉；

【作者基本信息】 西南交通大学 ， 电力系统及其自动化， 2009， 博士

【摘要】 行波故障测距法按所需故障信息可分为单端和双端实现方法,双端法不依赖行波在线路上的多次反射现象因此可靠性和计算精度较高,但同时它的成本较高。单端行波测距法由于仅需单端信息,不需时钟同步设备和通讯设备,与双端法相比在经济性上有相当大的优势,但现有的单端测距实时算法可靠性较差,因此单端行波测距法是行波测距的研究热点之一。本文提出了一种仅利用单端行波的频率信息进行精确故障定位的新方法。与传统的基于电报方程的行波分析方法不同,本论文把有损的均匀传输线等效为多端口网络,两端相连的电力系统等效为集总电路,然后把它们当作一个整体建立数学模型。在这个数学模型基础上推导输电线端点处的电流和电压在拉普拉斯域的函数表达式,然后由函数表达式的特征公式可以得到行波函数的极点与线路长度、边界条件三者的数学关系。另一方面,各个极点的位置(在本文中统一称为行波固有频率)与对行波波形进行频谱分析得到的各个频率峰值点一一对应。于是在线路参数已知的条件下,利用适当的频谱估计方法提取行波的固有频率(一般采用主成分),结合边界条件即输电线终端连接的系统等效阻抗即可精确计算形成固有频率的两个反射点之间的线路长度,对故障行波固有频率两个反射点之间(一端为本端母线,另一端为故障点)的线路长度就是故障距离。从理论上推导了单相或多相线路、无损或有损输电线、线路参数与频率无关或相关等多种情况下故障行波固有频率与线路长度、边界条件三者的数学关系,使理论基础完善。然后讨论了新算法的实现必须考虑的几方面问题:首先研究了算法在不同故障方式下的实现方法及其适用性。在此基础上对单相接地故障时的“模混杂”现象从理论分析和仿真计算两方面出发进行了分析。提出了提高计算精度的新算法。研究了对故障行波固有频率主成分进行频率估计的方法。对传统频域频率变换方法、时频域方法和基于谐波模型的频谱估计方法在提取故障行波固有频率主成分应用中的优缺点进行了对比分析。讨论了不同的母线结构对算法的影响。母线结构的变化将引起行波波头形状的变化和频谱上干扰频率的产生,前者不会对固有频率主成分法造成影响,在这点上本算法优于时域行波法;后者会对算法造成影响,文中提出了新的波头识别方法,通过该方法可以发现在频域仍然比时域上更容易区分故障点和对端母线的反射波。研究了互感器设备的频率传变特性对算法的影响。通过研究,提出了利用电容式电压互感器二次侧电压行波进行单端故障定位的新方法。在故障行波固有频率主成分测距法的理论和实现方案的各个方面都得到考虑的基础上进行了大量仿真试验对算法的精度进行检验。更多还原

【Abstract】 Travelling wave fault location can be classified into single-ended and double-ended methods according to different ways of obtaining the fault information. Double-ended method does not depend on multiple reflections of the travelling wave between the station buses, therefore it has higher reliability and accuracy. On the other hand, single-ended method relies on information from one end only. It does not need time synchronization or communication devices, thus more economical. However, the present single-ended methods are less reliable or accurate, which make single-ended travelling wave fault location one of the hottest subjects of research.A new method of accurate fault location using frequency information from one end only is presented, which is different from the traditional analyzing methods based on telegrapher’s equations. In this paper the lossy homogeneous transmission lines are described by the multi-port model, and the power systems connected at each ends of the line are represented by lumped circuits. The model is thus composed of transmission lines and lumped circuits. It is then used to derive the transmission line terminal voltages and current functions in Laplace domain. Using the characteristic equations of the terminal voltages or currents functions, the relations of poles of travelling wave, line length and boundary conditions are further derived. On the other hand, the poles of the voltage/current functions are identical to components of the frequency spectra (hereafter referred to as natural frequencies) resulted from frequency transform of the voltage/current signals. Given the line parameters, using adequate frequency estimation methods to obtain any component of travelling wave natural frequencies (normally the dominant component), together with the boundary conditions (power system equivalent reactance), the distance between two terminals which forms the natural frequencies can be accurately calculated. For fault induced travelling wave frequencies (local bus the one end, fault point the other), the length is naturally the fault distance.The relations of fault induced natural frequencies, fault distance and boundary conditions are thoroughly discussed at single phase and three phases, lossless and lossy lines, frequency independent and dependent line parameters. This helps to make the theoretic foundation of the algorithm complete.After that, various aspects of the realization of the new algorithm are discussed:First of all, the realizations of the new algorithm under different fault conditions are studied. Then the modal mixing phenomenon at single-phase-to-ground fault is analyzed. A novel algorithm which improves the accuracy of fault location under the modal mixing conditions is presented and discussed theoretically and through simulations.Different ways of estimating the fault induced travelling wave natural frequencies are discussed. The pros and cons of traditional frequency transform methods, time-frequency transformation and parametric spectral estimation to this particular application are compared and discussed.The effects of different bus configurations on the algorithm are discussed. The change of bus configurations may cause the variation of the shape of the travelling wave and the formation of additional natural frequencies. The new algorithm is immune to the variation of the wave shape, a property which is superior to traditional time-domain method. The formation of additional natural frequencies has an influence on the algorithm. New approaches of identifying the correct wave front are also introduced in this paper. It can be found out that it is still easier in the frequency domain than in the time domain to distinguish the wave reflected by fault point and the remote bus.The effects of transient response of instrument transformers to the algorithm are studied. A new single-ended fault location method using capacitive voltage transformers (CVT) secondary voltages is presented.After all aspects of the theories and realizations of the travelling wave natural frequencies fault location method are taken care of, a large amount of simulations are conducted in order to verify the accuracy of the algorithm.更多还原