电力系统继电保护快速滤波算法研究

【作者】 黄瀛；

【导师】 何奔腾；

【作者基本信息】 浙江大学 ， 电力系统及其自动化， 2005， 博士

【摘要】 传统继电保护滤波算法,由于其滤波速度与滤波能力间的矛盾,速度与准确度难以进一步提高。为解决传统滤波算法的问题,论文对高压输电线路的暂态过程、最小二乘算法、卡尔曼滤波算法、自适应滤波算法、非线性最小二乘算法、遗传算法理论以及算法误差评估理论进行了深入的研究,主要工作是寻找速度更快、准确度更高的保护滤波算法或算法配置策略。所取得的成果包括: 1.提出高压输电线路故障高频噪声信号“主频”理论,指出故障信号高频噪声分量中存在一个主频,其能量远远大于高频噪声中的其它频率成分,其频率随故障距离和系统阻抗的不同变化范围较大。 2.提出了一种基于频率—衰减时常特性的最小二乘算法模型性能评价方法,并使用此评价方法对最小二乘算法的各种模型进行了评价,得到了最小二乘算法的最佳模型。 3.提出了基于高频噪声主频估计的自适应最小二乘算法,大大提高了保护滤波算法的计算速度。 4.研究了非线性最小二乘算法在电力系统滤波计算中的应用问题。 5.提出了滤波算法误差评估器的概念和一种实现方案。 论文各章概貌简述如下 第一章对高压输电线路故障暂态过程进行了全面的研究,提出了故障高频噪声主频理论,并对故障信号暂态噪声各分量的频率成分及各频率成分间的关系进行了研究。第二章通过对电力系统常用的最小二乘算法、傅氏算法和卡尔曼滤波算法的研究指出,傅氏算法可以看作最小二乘算法的一种特例;而当把输入信号暂态噪声当作白噪声的情况下,卡尔曼滤波算法蜕化为最小二乘算法。第三章对最小二乘算法的各种模型进行了定性的分析和定量的比较,为此还提出了一种最小二乘算法模型性能定量比较方法。第四章通过对自适应滤波理论和遗传算法的研究,提出了基于高频噪声主频估计的自适应最小二乘算法。第五章针对基于高频噪声主频估计的自适应最小二乘算法计算量过大的问题,对非线性最小二乘算法在电力系统中的应用进行了研究,通过把高频噪声主频频率放入非线性最小二乘算法的待求量,提高了快速滤波算法的计算速度。第六章通过对滤波算法误差与输入信号的噪声水平间关系的研究,提出了误差评估器的思想和一种更多还原

【Abstract】 Because of the controdicition between the speed and performance of filter algorithms, the speed and performance of traditional filter algorithms couldn’t be improved furtherly. To solve the problem of traditional filter algorithms, the thesis does comprehensive research on the transient process of high voltage transmission line, least error square algorithm(LES), Kalman algorithm, adaptive algorithm, non-linear LES, genetic algorithm and algorithm error evaluation theory. The main work lay in searching more rapid and accurate algorithm or algorithm deployment policy. The results are:1. The thesis proposes fault signal high frequency components "main frequency" theory , that indicates there is a "main frequency" component in fault signal high frequency components, its energy is larger than other components of high frequency components significantly, its freuqency is determined by fault distance and system impedance, and ranges widely.2. The thesis proposes a filter frequency-decay constant property based LES model performance evaluation method, and uses this method to evaluate all kinds of model of LES to get the best model.3. The thesis proposes a high frequency components "main freqency" evaluation based LES algorithm, the algorithm improves the speed and accuracy of relay protection filter algorithm significantly.4. The thesis does research on the application of non-linear LES in power system relay protection filter computing.5. The thesis proposes the concept of filter algorithm error evaluation and an implementation based on LES.The skeleton of each chapter followsChapter one does comprehensive research on the transient process of high voltage transmission line and proposes the fault signal high frequency components "main frequency" thoery. On the other hand, the thesis does some research on the constitution of frequency components and the relation among them. Chapter two indicates that Fourier algorithm can be seen as a particular case of LES algorithm and in some circumstance Kalmanalgorithm becomes LES algorithm by the research on common LES algorithm, Fourier algorithm and Kalman algorithm. Chapter three analyses all kinds of models of LES algorithm qualitatively, compares their performance quantitively. and proposes a performance evaluation method for that. Chapter four proposes high frequency components "main frequency" evaluation based adaptive LES algorithm by the research on the adpative filter thoery and genetic algorithm. Chapter five does some research on the application of non-linear LES algorithm in power system relay filter computing to solve the computing speed problem of adaptive LES algorithm proposed in prev chapter. Chapter six proposes the concept of filter algorithm error evaluation and an implementation based on LES algorithm though the research on the relation between the erorr of filter algorithm and the noise of input signal. The chapter proposes an relay protection scheme based on filter algorithm error evaluation too. Chapter seven develops an Linux cluster based simulation system to satisfy the demand of great mount simulation computing in relay protection research. Chapter eight is the conclusion of the thesis.更多还原