DFT analysis of continuous non-periodic signal considerations

1. The sampling frequency has a greater effect on the frequency spectrum of the DFT analysis signal because it directly affects the degree of spectral aliasing. The sampling frequency must be greater than twice times the maximum frequency of the signal.

2. The fence phenomenon is related to the resolution of frequency, because the discrete Fourier transform is the Fourier time transformation between [0,2pi] interval sampling, when the sampling points are small, it can not reflect all the details, leading to the fence phenomenon. One way to overcome this is to make a sequence of length l>n, and then perform a DFT operation, by filling 0 after the truncated length of N. 0 can only improve the observation resolution, but can not increase the information, thereby improving the distortion.

3. Spectral leakage: Due to the addition of windows in the time domain signal, it is necessary to select the appropriate window function. The addition of Windows has two effects on signal spectrum analysis: First, there are redundant high-frequency components in the spectrum, and secondly, the spectral line becomes a spectral peak with a certain width, thus reducing the resolution of the frequency.

The spectral aliasing is related to the sampling interval of continuous signals, and the spectral leakage is related to the length of time domain plus window truncation, and the fence effect is related to the number of DfT points. Replication Search Replication Search replication Search replication Search replication Searches copy Search