频谱校正中频率-幅值估计的一种新方法

指出了导致频谱误差的根本原因是标准DFT计算中的频率初始点为零,并由此提出一类新的频谱校正方法——初始频率扫描法,包括单频率估计方法和全频段估计方法.该方法以幅值最大为判据准则,以一个频率分辨力为寻优区间,通过调整DFT计算的初始频率点实现对频率和幅值的最优估计.验证结果表明,该方法估计精度较高,能够稳定保持在10-3数量级,且估计精度对信号点数的要求不高,同时,噪声的强烈程度对估计精度影响并不显著,说明该方法具有很强的实用性.

A Novel Method for Estimating Frequency and Amplitude in Spectrum Correction

It is unavoidable that considerable errors will be found in the results of the discrete Fourier transform(DFT),of which the root cause is that the initial frequency is set to zero in DFT calculation.A novel method named initial frequency scan(IFS) is therefore proposed to estimate the frequency and amplitude for spectrum correction,including both the single and whole frequency band estimation. The basic principle of IFS is to scan different frequencies in a frequency resolution as optimization interval to estimate the frequency and amplitude with the max amplitude as criterion.Then,an optimal frequency/amplitude estimation is implemented via the initial frequency point that is found by regulating DFT calculation.The results show the high accuracy of IFS,i.e.,the error magnitude can be kept at 10-3,without strict requirement of signal units.In addition,the noise intensity affects unobviously the estimation accuracy.All these verify that the method proposed is highly practical.