【作者】 刘炳旭；

【导师】 赵建国；

【作者基本信息】 山东大学 ， 电力系统及其自动化， 2007， 博士

【摘要】 随着广域测量系统(WAMS)在电力系统中的推广应用，基于广域信息的电网监控新技术的研究不断开展，电网监控技术将逐步进入实时动态监控的阶段。基于全球定位系统(GPS)提供的同步信息对电网中的相量进行精确测量是同步相量测量单元(PMU)的基本功能，是WAMS系统各项功能实现的基础。PMU首先要对原始电气信号进行同步采样，计算出带有时标的相量信息。离散傅立叶变换(DFT)方法是计算相量的基本算法，当电网频率偏移时，由于栅栏和频谱泄露现象，相量的计算将出现误差。根据系统实际频率的变化来调整采样频率，可以实现信号整周期采样，从而减小误差，但WAMS对相量同步性的要求，使该方法的应用受到限制。相量有旋转相量和固定相量两种表达方式，固定相量的相角为实际相量与工频相量之间的相对相角，二者只是表达方式的不同。两种方式下，DFT法与递归DFT法有不同的运算表达式。相量具有时间性，DFT方法的计算相量为数据窗中参考点的时刻。此外，参考点的选取还影响着DFT方法的计算结果，当以数据窗中心点为参考点时，频率偏移情况下相量的计算误差最小。通过对连续傅立叶变换计算结果的研究，提出可以对DFT的相量计算结果进行校正得到精确的相量值。实际的仿真计算结果表明，在稳态条件下，对幅值不变，频率偏移额定值的信号，利用定间隔采样的结果，该校正计算方法可以实现对相量的精确计算。而频率变化缓慢时，测量误差较小，完全能满足实际系统的需要。同时利用校正系数可以检测电网的运行状态，校正系数稳定不变表明信号为单一频率的稳态信号；校正系数变化则说明测量信号波动或发生突变。在电力系统低频振荡条件下，基于DFT方法的相量校正算法将不再适用。理想的低频振荡信号可以看作是基波信号的幅值以低频正弦规律变化，对常规DFT算法在理想低频振荡下的计算结果的分析表明，其计算相量可以作为计算低频振荡参数的依据。研究了利用小波变换计算低频振荡信号的方法，并以Morlet复小波和改进递归小波为例，对理想条件下DFT方法的计算结果和动模实验中PMU的测试数据进行了分析，通过相位计算测量出低频信号的频率。利用基于DFT方法的相量计算的基本原理，以固定采样频率设计实现了Windows环境下的同步相量测量装置。分析了PMU系统应具备的性能指标与Windows操作系统在实时性方面的不足，并利用多线程与VxD技术对软件的运行机制进行了改进，有效提高了系统运行的实时性。在动模实验室环境下进行了稳态、短路、振荡等条件下的测试实验，并以实验数据基础进行了大量仿真计算。更多还原

【Abstract】 With the wide application of Wide-Area Measurement System (WAMS) in power system, the research of new technology about monitoring power networks is developing greatly. It is the basic function of Phasor Measurements Unit (PMU) to measure precisely the phase of network by the synchronized data from Global Positioning System (GPS), and it is the base of other functions of WAMS.First, PMU will sample the initial signal, and calculate the phase with time tag. Discrete Fourier Transforms (DFT) is the basic method to calculate phase. It will result in the error of calculating phase when the real frequency excursion. If the sampling frequency is regulated as the frequency of network, it can sample at a full period and the error will be reduced. But because of the requirement to synchronization of WAMS, the method is limited.There is two styles of phase , rotating and const phase . The angle of const phase is the difference of real angle and the one with rated frequency. Under the two style, DFT and recursion DFT have different formula.The phase is relative to time. The time of the phase from DFT is relative to the reference point of the data window. Besides, the selection of reference point affects the precision of phase calculating by DFT. When the point is the middle of data window, the error is the least. After reseaching the result of calculating phase by DFT, a new phase revising method is raised. It can be known from the simulated calculteing, that the mothed can calculate the phase precisely when its ample is const and frequency varys from its rated value under statbility of power system.Under the condition of low frequency oscillation, the revising method of phase based on DFT is not suitable any longer. The theoretic signal of oscillation can be regarded as the principal wave whose ample changes as sine curve. It is illustrated by the calculating that the result of phase calculating by DFT can be as the data to anylise low frequency oscillation. After summurising the mothed of calculating oscillation by wavelet transform, this thesis calculated the main frequency of theoretic occilation signal.Based on the principle of calculating phase by DFT, a PMU is designed and implemented under the Windows system, using the fixed sampling frequency. After analyzing the performance index of PMU, this dissertation discusses the Windows system’s character and its shortage in the real time performance. Virtual Device Driver (VxD) and multithread are used in the software developing, which have enhanced the real time performance greatly.更多还原