一种改进的插值FFT谐波分析算法  

An Improved Interpolated FFT Algorithm for Harmonic Analysis

作　　者：翟晓军[1] 周波[1] 

机构地区：[1]江苏省新能源发电与电能变换重点实验室(南京航空航天大学),江苏省南京市210016

出　　处：《中国电机工程学报》2016年第11期2952-2958,共7页Proceedings of the CSEE

摘　　要：快速傅里叶变换(fast Fourier transform,FFT)是进行谐波分析的有效方法,能够在很大程度上减小计算量。进行FFT计算存在栅栏效应导致观测误差,采用插值算法能够对误差进行修正。进行FFT计算的数据长度必须满足一定条件,不利于采样参数的灵活设置。该文以基2的FFT算法为例,针对采样数据长度非基2的场合,采用补零的方法使进行FFT计算的数据长度满足基2条件,分析给出了补零后的插值算法修正公式。分析结果表明,当采样数据非基2的条件下,采用该文方法进行FFT插值计算与离散傅里叶变换(discrete Fourier transform,DFT)插值计算是等效的,且使采样参数设置更加灵活,不需满足基2的条件。以飞机400 Hz交流供电系统为例,通过仿真和实验证实了该方法的可行性和有效性。Fast Fourier transform（FFT） is an effective analysis method for harmonics, and reduce the computational complexity to a great extent. FFT has the picket fence effect which leads to the observation error. The observation error from picket fence effect can be revised by the interpolated algorithm. However, the number of sampling data must satisfy certain conditions, which is not conducive to flexible set of sampling data. When the number of the sampling data is non radix-2, the paper calculated FFT through zero padding method to satisfy radix-2, and analyzed theoretically the interpolated FFT algorithm formula with the zero padding. The analysis results show that when the sampling data is non radix-2, the interpolated FFT algorithm and the interpolated discrete Fourier transform（DFT） algorithm are equivalent. It is more flexible in sampling data set, and does not need to satisfy the radix-2 conditions. Combining with the applications of 400 Hz AC power supply system, the paper validate the practicability and effectiveness of the algorithm through the simulation and experimental results.

关 键 词：频率估计 栅栏效应 快速傅里叶变换 插值算法 谐波分析 

分 类 号：TM935[电气工程—电力电子与电力传动]