基于频差修正的余弦信号频率估计算法

针对余弦振动信号的频率高精度估计需求,提出了一种基于频差修正的频率估计算法.对连续时间信号进行采样后,使用Candan算法估计出频差,运用频差对信号的频率进行修正.对修正后的信号使用Liang算法再次进行频偏估计.最后将2次估计得到的频率值相加求得最终估计频率.通过频差修正,避免了Candan算法因插值方向错误和Liang算法自身特点导致估计精度降低的问题,虽然增加了计算量,但并不影响信号实时处理.仿真结果表明:在相对频偏为任意值的情况下,改进算法的频率估计均方误差接近克拉美罗下限(CRLB),性能优于现有频率估计算法.

Frequency estimation algorithm for cosine signal using frequency-offset correction

To meet the demand of high precision frequency estimation of cosine vibration signal,a new algorithm based on frequency-offset correction is proposed.The algorithm estimates the frequency-offset,with which the frequency of signal can be revised,by Candan algorithm after sampling the continuous time signal.For the signal revised,use Liang algorithm to estimate the frequency-offset again.The ultimate estimated frequency is given by adding the two estimated frequency together.By frequency-offset revision,the proposed algorithm avoids the decrease of precision caused by interpolation error of Candan algorithm and characteristics of Liang algorithm.The new algorithm guarantees the real-time signal processing,although it increases the amount of computation.Numerical results demonstrate that the mean square error of frequency estimation of the improved algorithm is more closer to the Cramer-Rao lower bound(CRLB) bound character is prior to the existing frequency estimation algorithm.