摘要
提出了基于快速傅里叶变换与误差最小原理的电力系统谐波分析方法。该方法设定了畸变波形模型和修正参数,当模型波形与实际波形之间的均方差最小时,模型参数可以代表实际波形的参数。为避免分析出的谐波次数不准确而出现无效参数,把实际采样数据分成训练组和测试组。在训练组中通过最陡下降梯度查询学习策略的迭代循环修正参数。在测试组中检测谐波次数的正确性,获得有效的谐波分析结果,实现对次谐波和频率相隔很小的谐波的同步跟踪。仿真实验验证了该方法的有效性与易实现性。
A method to analyze harmonics in power system based on fast Fourier transform（FFT） and minimal error principle is presented, in which a model of distorted waveform and correction parameter are set up, the parameters of the model can represent the parameters of actual waveform while the mean square deviation between the waveform of the model and actual waveform is minimal. At the same time, to avoid invalid parameters caused by inaccurate order of harmonics being analyzed, the actual sampled data is divided into two groups, i.e., training group and testing group. In training group the correction parameters of iterative loops for search strategy can be inquired about; in testing group the correctness of harmonic order can be detected to obtain effective results of harmonic analysis, so the synchronous tracking of non-integer harmonics and harmonics with small frequency interval can be realized. Simulation results show that the proposed method is effective and easy to implement.
出处
《电网技术》
EI
CSCD
北大核心
2006年第19期76-79,共4页
Power System Technology
关键词
快速傅里叶变换(FFT)
参数修正
均方差
电力系统
谐波
fast Fourier transform（FFT）
parametric modification： mean-squared error
power system
harmonics
