基于FICP算法提高对低频噪声时延的估计精度

针对非平稳随机噪声（直升机飞行声音，导弹喷流噪声）信号，研究了提高时延估计计算精度的方法，通过计算信号的细化频谱，并避免了标准FFT计算带来的栅栏效应，从而使得对于超短取样信号的频谱计算精度与信号长度无关，采用快速计算方法FICP由N点互谱计算2N点相关波形，提高了相关波形的分辨率，使得时延估计精度得以提高。对实测直升机飞行噪声和导弹噪声计算了不同取样长度和不同采样率下的时延估计，得出了谱分辨率越高，时延估计精度越高的结论。

Improving the accuracy of time-delay estimation for low-frequency noise based on the algorithm of FICP

A new way has been studied to improve accuracy of time-delay estimation (TDE) as for non-stable random signal such as noise of helicopter, missile. A barrier affection caused by standard FFT calculation was reduced by calculating signal's fine spectrum. Calculating accuracy of signal's spectrum will have no relationship with length of the sampling signal. A new fast algorithm of the transform FICP (Fine Interpolation of Correlation Peak) was used to calculate a 2N sequence of correlation waveform in time domain from an N sequence in frequency domain, in order that the accuracy of TDE can be improved after the accuracy of correlation waveform being increased. As for noise of the helicopter detected in actual flying and of the missile, TDE have been calculated with different sample length in different sampling rate. A result has been obtained that the accuracy of TDE will be higher when cross spectrum is more accurate.