【作者】 苏玉香；

【导师】 刘志刚；

【作者基本信息】 西南交通大学 ， 电力系统及其自动化， 2008， 硕士

【摘要】 近几年来,由于电力电子器件和非线性装置的广泛应用,使得电网中的电压和电流波形畸变越来越严重,造成了电能质量的恶化。传统的电能质量检测常用傅立叶变换方法,但该方法不适用于非平稳、非线性的暂态电能质量检测。因此寻求新的更适用于非平稳、非线性信号的分析方法就显得尤为重要。本文利用Hilbert-Huang变换方法对电能质量进行检测分析,着重分析了电力系统短时电能质量扰动信号及谐波信号,并且首次将该方法用于电气化铁道谐波电压、谐波电流实测数据的检测分析中。通过对实测数据的仿真分析,结果表明Hilbert-Huang方法在电气化铁道谐波检测中是非常有效的。本文的主要工作如下:深入研究了Hilbert-Huang变换理论,并着重研究如何运用该方法检测电网电能质量问题,包括暂态电能质量(电压骤升、电压骤降、电压中断,暂态振荡和暂态冲击等)的检测和谐波(稳态谐波、间谐波、分数谐波和时变谐波等)检测。针对HHT方法中存在的端点效应问题,本文在原有改进方法的基础上,提出了两种改善方法:基于人工神经网络和镜像延拓相结合的新的时间序列延拓方法、基于支持向量回归机和镜像延拓相结合的新的时间序列延拓方法,这两种智能方法与镜像延拓方法的结合,取长补短,仿真分析结果表明其改善端点效应的效果非常好。本文还对所提的两种方法作了比较分析,结果表明,对于大多数信号而言,基于支持向量回归机和镜像延拓相结合的方法要优于基于人工神经网络和镜像延拓相结合的延拓方法。本文详细分析了电气化铁道谐波的特点,在此基础上综述了目前电气化铁道谐波检测的常用方法及其各自的优缺点。由于电气化铁道电压、电流信号中基波的能量相对其它各次谐波的能量大得多,直接应用HHT方法存在模态混叠问题,不能将基波以及各次谐波有效地分开;因此本文将Yang提出的基于Fourier变换的EMD方法,并结合本文提出的基于支持向量回归机和镜像延拓相结合的数据延拓方法,应用于电气化铁道谐波检测中。应用该方法有效地分离出了基波及各次谐波信号。对分离出的单分量谐波信号进行Hilbert变换,可以得到各次谐波的瞬时频率和瞬时幅值,即可以得到真正意义上的时频分布。更多还原

【Abstract】 With the pervasion of power electronics apparatus and nonlinear loads, recently the power quality problems have been more and more deteriorated by the distorted waveform of voltages and currents in power network. Traditional power quality analysis methed is Fourier transform. As a traditional analysis tool in frequency domain, it is not suitable for the non-stationary and non-linear signals processing. Thus new analysis methods are required to detect and analysis power quality disturbances accurately.Hilbert-Huang Transform (HHT) is applied in detection and analysis of power quality in this paper. Short time power quality disturbances and harmotics are taken more consideration. For the first time, Hilbert-Huang Transform is applied to detect harmonic in traction power systemin this paper. And simulation results show that this method is effective to extract harmonic. The main contribution is gaven as follows:Hilbert-Huang Transform is studied deeply and it is applied to power quality in power system, such as short time power qulity disturbances including voltage sags, voltage swells, interruption, transient oscillasion and transient interruption, and harmonics including integer harmonic, non-integer harmonic, transient harmonic distortion, and so on.In order to restrain the end effects of Hilbert-Huang transform, two new methods, combination of support vector regression machines and mirrorizing extension, combination of BP neutral network and mirrorizing extension, are proposed in this paper based on the original extension methods. The merits of support vector regression machines, BP neutral network and mirrorizing extension are discussed, and the disadvantages of them are listed. Simulation results show that the two methods are effective to improve the end effects of HHT. Comparision of the two methods is given in this paper, and the results show that the former new method is better than the latter for most of simulation and real signals. The charateritsic of harmonics in traction power system is introduced and analyzed, and the corresponding applied methods for measuring harmonics are analyzed and discussed in this paper.Because the energy of fundamental is much bigger than the other harmonics’ in traction power system, mode-mixing appears seriously during the sifting by Empirical Mode Decomposition (EMD). Harmonic with any frequency in traction power system can not be directly abstracted with EMD. In order to sift the mode-mixing harmonic signal into good mode, an improved method named EMD based on Fourier transform, which is proposed by Yang, is used in this paper. This method is with the combination of support vector regression machines and mirrorizing extension method. The improved HHT method is applied to harmonics analysis in traction power system, and it is effective to extract harmonic with any frequency. Accurate instantaneous frequency and instantaneous amplitude of harmonic components can be obtained by using Hilbert transform.更多还原