一种基于非同步采样的FFT算法(英文)

A NOVEL FFT ALGORITHM FOR NONSYNCHRONOUS SAMPLE

在非同步采样中由FFT引入的频谱泄漏在谐波测量中会造成很大误差。以往采用加窗和插值的方法来处理这样的问题,但是,为了更精确地计算谐波,这些技术必须要延长数据采样长度来抑制谐波间的相互干扰。文中在详细分析了频谱泄漏机理的基础上,导出了信号的实际频谱和泄漏频谱之间的简明关系。该方法能用相对比较短的采样长度从FFT计算的泄漏频谱中获得真实频谱,并不需要考虑频谱泄漏和谐波间的相互干扰。仿真表明该算法比FFT精度更高,可很好地恢复出信号的真实频谱。

The leakage effect is introduced by Fast Fourier Transform (FFT) in nonsynchronous sampling, which may cause great errors in measuring harmonic components. Many literatures had been issued to cope with the problem using weighted windows and interpolations. However these techniques have to prolong the sample duration to avoid the mutual harmonic interference. Based on the intensive analysis of the leakage effect, a simple formula is deduced between the leaked spectrum and actual spectrum. The method retrieves the actual spectrum from the spectrum calculated by FFT with relatively short sample duration without considering the leakage effect and mutual interference among harmonics. Simulation results show the algorithm has higher accuracy than FFT and almost reconstructs the actual spectrum.