基于加窗插值FFT和原子分解的间谐波检测算法

发布时间：2018-04-01 05:26

本文选题：间谐波　切入点：加窗插值算法　出处：《郑州大学》2017年硕士论文

【摘要】：近年来,随着电力电子设备的大量使用和冲击性负荷的增多,电力系统中间谐波的含量越来越大。除了具有整数次谐波的危害外,间谐波还会造成电压闪变及零点偏移等问题,给电能质量带来了严重影响,威胁着电力系统的安全经济运行,因此治理间谐波问题迫在眉睫。只有准确检测出间谐波参数,分析出间谐波信号,才能对其进行抑制和治理,进而改善供电环境,确保供电设备正常运转。本文首先分析了傅里叶变换的理论和基本形式,FFT因为算法简单计算量小是检测谐波间谐波的常用方法。但当采样不同步时,FFT分析会出现频谱泄露、栅栏效应等问题,致使检测误差增大。针对这些问题,学者们提出了使用加窗插值算法来提高检测精度。但是,由于受到频率分辨率的限制,对于检测频率相近信号,加窗插值并不适用。然后,文章提出了使用原子分解算法对谐波间谐波信号进行检测。原子分解的两大核心问题是如何构建原子库和如何寻找最佳匹配原子,针对稀疏分解中常用的Gabor原子与间谐波的信号特征匹配度不高,提出了使用正弦量原子构建过完备原子库;由于使用匹配追踪算法寻找最佳原子时,计算量很大,对此对离散参数进行优化以减少计算量。经过改进和优化,使得原子分解算法能很好适用于谐波间谐分析检测中,为本文算法奠定了理论基础。最后综合加窗插值和原子分解两种算法的优势,扬长避短,提出了混合算法,并对算法的实现步骤进行了详细阐述。对用Hanning窗截断的信号进行FFT分析,求得真实频谱附近的三条谱线幅值;利用谱线干涉判据判断否存在谱线干涉;根据判断结果选择检测间谐波的算法。仿真表明该算法既减小了原子分解算法的计算量,又弥补了加窗插值分辨率的不足,即当谐波间谐波的频率间隔很近时,该算法也能获取间谐波参数。
[Abstract]:In recent years, with the large use of power electronic equipment and the increase of impact load, the content of intermediate harmonics in power system is increasing.In addition to the harm of integer harmonics, interharmonics will also cause voltage flicker and zero offset, which will seriously affect the power quality and threaten the safe and economic operation of power system, so it is urgent to solve the problem of inter-harmonic.Only by accurately detecting interharmonic parameters and analyzing interharmonic signals can it be suppressed and treated and the power supply environment can be improved to ensure the normal operation of power supply equipment.This paper first analyzes the theory and basic form of Fourier transform (FFT) because the simple calculation of the algorithm is a common method for detecting harmonics between harmonics.But when the sampling is out of sync, the spectrum leakage and fence effect will appear in FFT analysis, which will cause the detection error to increase.To solve these problems, scholars proposed a windowed interpolation algorithm to improve the detection accuracy.However, due to the limitation of frequency resolution, windowed interpolation is not suitable for detecting similar frequency signals.Then, an atomic decomposition algorithm is proposed to detect the harmonic signal between harmonics.The two core problems of atomic decomposition are how to construct atomic library and how to find the best matching atom. The signal characteristic matching degree between Gabor atom and interharmonic is not high in sparse decomposition.The over complete atomic library is constructed by using sinusoidal atom, and the discrete parameters are optimized in order to reduce the amount of computation because the matching tracing algorithm is used to find the best atom.Through the improvement and optimization, the atomic decomposition algorithm can be well applied to harmonic analysis and detection, which lays a theoretical foundation for the algorithm in this paper.Finally, combining the advantages of windowed interpolation and atomic decomposition, a hybrid algorithm is proposed, and the implementation steps of the algorithm are described in detail.Based on the FFT analysis of the signal truncated by the Hanning window, the amplitude of three spectral lines near the true spectrum is obtained; the spectral line interference criterion is used to judge whether there is spectral interference; and the algorithm for detecting interharmonic is selected according to the result of the judgment.Simulation results show that the algorithm not only reduces the computational complexity of the atomic decomposition algorithm, but also makes up for the lack of windowed interpolation resolution, that is, when the frequency interval between harmonics is very close, the algorithm can also obtain the inter-harmonic parameters.
【学位授予单位】：郑州大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM935

【相似文献】