基于加窗插值的高精度谐波检测算法研究

发布时间：2018-05-23 06:30

本文选题：谐波检测 + 高精度　；参考：《中国矿业大学》2017年硕士论文

【摘要】：电网中的谐波污染日渐严重,造成了电压电流波形的畸变,威胁了电网的安全运行。对畸变电压电流中各次谐波参数的准确检测是精准化治理的重要前提。本文围绕着高精度的加窗插值谐波检测算法展开了研究。首先,通过分析现有几种谐波检测算法的优缺点,确定快速傅里叶变换(Fast Fourier Transform,FFT)算法是谐波检测中最适用的算法。其次,对FFT算法中的两个主要问题进行了研究。第一个问题是由于截断采样信号引起的频谱泄露,分析了出现频谱泄露的现象及原因,研究了能有效抑制频谱泄露的窗函数,分析了窗函数的选择标准并研究了几种常见的窗函数;第二个问题是由于非同步采样引起的栅栏效应,分析了栅栏效应的现象,研究了能减少栅栏效应误差的插值算法,并利用MATLAB中的cftool工具箱针对不同的插值算法计算出常用窗函数的修正公式。再综合两个解决方法研究出完整的加窗插值FFT算法。再次,提出基于五项最大旁瓣衰减(Maximum-Sidelobe-Decay,MSD)窗三谱线插值FFT的谐波检测算法。分析了五项MSD窗的时域和频域特性,结合三谱线插值算法形成完整的谐波检测算法。通过MATLAB平台编写算法程序,对比仿真该算法和其他加窗插值算法在多种情况下对谐波的检测效果,结果显示该算法在所有谐波环境中都具有特别高的检测精度,同时该算法能够有效观察信号的全频谱信息,方便对信号中的其他成分进行进一步分析。随后,提出基于五项MSD窗三谱线插值改进Goertzel的谐波检测算法。研究了Goertzel算法并对其进行了改进,结合五项MSD窗三谱线插值算法形成完整的谐波检测算法。研究了该算法实现谐波检测的全过程,通过只计算三谱线插值算法中所需的三条谱线的谱值来节省计算时间。通过MATLAB平台,对比仿真该算法和五项MSD窗三谱线插值FFT算法对谐波的检测效果,结果显示该算法具有同样高的检测精度,在无法观察信号全频谱信息的情况下提升了计算速度,特别适用于对检测精度和实时性要求都很高的常规谐波环境中。最后,搭建实验平台,得到待分析的实验谐波离散信号。通过本文的两种算法和其他算法对实验信号进行检测。结果显示,本文提出的两种算法都有很高的检测精度,而且在检测功能和速度上各有特色。
[Abstract]:The harmonic pollution is becoming more and more serious, which results in the distortion of voltage and current waveform and threatens the safe operation of power grid. Accurate detection of harmonic parameters in distorted voltage and current is an important prerequisite for precision treatment. This paper focuses on the high-precision windowed interpolation harmonic detection algorithm. Firstly, by analyzing the advantages and disadvantages of several existing harmonic detection algorithms, it is determined that Fast Fourier transform algorithm is the most suitable algorithm in harmonic detection. Secondly, two main problems in FFT algorithm are studied. The first problem is spectrum leakage caused by truncation of sampling signal. The phenomena and causes of spectrum leakage are analyzed, and the window function which can effectively suppress spectrum leakage is studied. The selection criteria of window function are analyzed and several common window functions are studied. The second problem is the fence effect caused by asynchronous sampling, the phenomenon of fence effect is analyzed, and the interpolation algorithm which can reduce the error of fence effect is studied. By using the cftool toolbox in MATLAB, the modified formula of window function is calculated for different interpolation algorithms. Then two solutions are synthesized to develop a complete windowed interpolation FFT algorithm. Thirdly, a harmonic detection algorithm based on five-term maximum sidelobe attenuation Maximum-Sidelobe-Decay-MSD window three-line interpolation FFT is proposed. The characteristics of five MSD windows in time domain and frequency domain are analyzed, and a complete harmonic detection algorithm is formed by combining the three spectral line interpolation algorithm. The algorithm program is written on the MATLAB platform, and compared with other windowed interpolation algorithms, the results show that the algorithm has a very high detection accuracy in all harmonic environments. At the same time, the algorithm can effectively observe the full spectrum information of the signal and facilitate the further analysis of the other components of the signal. Then, an improved Goertzel harmonic detection algorithm based on five MSD window and three spectral line interpolation is proposed. The Goertzel algorithm is studied and improved, and a complete harmonic detection algorithm is formed by combining the five-item MSD window three-line interpolation algorithm. The whole process of harmonic detection by this algorithm is studied. The calculation time is saved by only calculating the spectral value of three spectral lines needed in the interpolation algorithm of three spectral lines. By using MATLAB platform, the harmonic detection effect of this algorithm and five MSD window three-line interpolation FFT algorithm is compared. The results show that the algorithm has the same high detection accuracy and improves the calculation speed without observing the full spectrum information of the signal. It is especially suitable for conventional harmonic environment with high detection accuracy and real-time requirements. Finally, the experimental platform is built to get the experimental harmonic discrete signal to be analyzed. Two algorithms and other algorithms are used to detect the experimental signals. The results show that the two algorithms proposed in this paper have high detection accuracy and have their own characteristics in detection function and speed.
【学位授予单位】：中国矿业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM935

【参考文献】