卷积检测:一种新FFT的故障检测机制

CONVOLUTION DETECTION:A NEW MECHANISM TO DETECT FAULT FOR FFT

首先分析ＦＦＴ阵列中蝶形单元的故障模式，提出一致性故障的概念，并根据傅立叶变换的性质，提出一个新的故障检测机制：卷积检测，并证明了该机制对单输出端故障和一致性故障的覆盖率为１００％ ．卷积检测机制的变换方法灵活，且具有实时性．符合实际工程应用的需要．在文章的最后。

The Fast Fourier Transform is a major analytical tool in many research fields. A new mechanism to detect fault for FFT is presented in this paper. The fault mode for the butterfly element in the FFT butterfly array is analyzed first, and the concept of consistent fault is first presented. A new fault detection for FFT, convolution detection mechanism, is proposed in this paper and is proved to be complete for single output fault and consistent fault. The convolution detection is applicable in real engineering situation because it is feasible and real time. Finally, the round off error is analyzed with the mathematical model.