【作者】 徐志向；

【导师】 吕厚余；

【作者基本信息】 重庆大学 ， 电气工程， 2006， 硕士

【摘要】 电力系统谐波已经成为影响电能质量的一个重要方面,实际的检测和治理需要对电网中的谐波源进行识别和定位。本文采用谐波状态估计方法先得到整个电网谐波分布的估计值,然后用估计结果判断实际谐波源的位置;结合监视控制和数据收集(SCADA—Supervisory Control and Data Acquisition)系统,为实现电网谐波的实时监控提供了基础。现代量测技术的发展为谐波状态估计技术提供了前提条件,同时带来了量测量、估计模型、求解算法以及可观性分析等相关问题的改变。本文首先介绍了PMU(Phasor Measurement Unit)量测技术基本理论,选择母线电压、母线注入电流和支路电流同步量测作为量测量,整个电网的母线电压向量作为状态量,接着介绍了量测的数学模型和估计方法,然后提出了相应的求解算法以及在此量测配置的优化方案,通过模拟仿真和程序进行验证;最后介绍了不良数据的检测和状态估计结果评估的方法。本文的主要工作及结论如下:①对于有足够量测方程(超定)的估计求解算法,当量测方程无病态出现时,可以直接通过一般的矩阵运算求解得到估计值。本文通过节点编号优化,运用分层算法对量测矩阵进行预先处理后再进行矩阵求解;②对于量测方程病态时,可用正交变换法、混合法以及奇异值分解算法(SVD—Singular Value Decomposition)求解。本文用运算简单快捷的SVD算法求解病态时的量测方程;③对于量测方程欠定且量测矩阵病态时的量测配置下,量测方程数不够,本文首次将SVD算法运用于估计方程欠定时的谐波状态估计问题,求得估计方程的最小二乘解;④首次在SVD算法的基础上分析部分可观系统的量测问题,对量测配置进行优化。通过相应的算例和程序运行结果表明:①优化节点编号可以明显简化量测矩阵结构,分层算法把整个系统的状态估计分割为单母线的状态估计,减少了矩阵运算量;②SVD分解算法对于病态的量测矩阵以及量测方程欠定的情况可以有效的求解,其解的精度较高,能够准确的对谐波源进行识别和定位;③在SVD分解的基础上的优化程序结果表明在量测量大大减少的情况下,仍然可以得到精度较高的估计结果,根据优化可以得到有效估计的最优量测配置方案,从而改善了基于PMU量测状态估计的经济效益。更多还原

【Abstract】 Harmonic becomes an important element of impact for electric quality in power system, harmonic sources identifying and location are necessary for actual detection and management. The text firstly gets the harmonic distribution of the entire electric network with harmonic state estimation (HSE) method, it is enough to locate the harmonic sources. It paves the way to monitor harmonic in real time together with SCADA ( Supervisory Control and Data Acquisition ) system.The modern measurement technology is prior condition of the HSE , meanwhile it brings some advancement for the relevant problems about measurement variables、estimation models、solution algorithm and observation analysis. Firstly the text introduces the basic theory of Phasor Measurement Unit (PMU), and selects the synchronism measurement value of bus voltage、bus influx current and branch current as measurement variables, the bus voltage in the whole net as status variables,the following brings up measurement models and state estimation methods, then advances homologous solution algorithm , brings forward optimization allocation project on basis of it. The simulation and procedures are used to verify it. Last of all, detection methods of the bad data and evaluation for HSE are proposed. The major work and conclusion as follows:①For enough measurement equation (overdetermined), estimation values can be gotten directly through simple matrix calculation when measurement matrix is without illness. The article deals with the measurement matrix in advance with laminarization algorithm by node numbering optimization.②For illness measurement matrix, there are some ways to solute equations such as orthotropic transformation method、hybrid method and SVD method, the SVD is used to solve illness equation due to its simple and fast calculation.③For underdetermined measurement equations and illness measurement matrix, there are not enough equation quantities. SVD is used to solve harmonic state estimation firstly, then the least square solution can be obtained by SVD.④Building on SVD method, the thesis analyzes measurement allocation problem of part observable system firstly. Go further, measurement allocation can be optimized.The computational results of the examples and procedure show that:①Measurement matrix is simplified by node numbering optimization. With laminarization algorithm, HSE of the whole net can be divided into HSE of the single bus,更多还原