经验模态分解在相关峰检测和测向中的应用

互相关测向法在低信噪比时检测不到相关峰,无法实现有效的方位估计.经验模态分解可以在突出信号局部特征的基础上,把信号分解成多个固有模态函数.采取对互相关函数进行经验模态分解,适当选取分解出的IMF重构互相关函数.重构的互相关函数有明显的相关峰,可以有效地提高检测目标的能力.将经验模态分解与互相关测向法相结合,可以实现低信噪比下的时延估计,从而更好地估计目标的方位.仿真结果表明重构的互相关函数较原互相关函数,其相关峰较最高伪峰的相对值提高了约4.9 dB,从而可以更好地实现时延估计.方位估计结果显示此方法在信噪比为-14 dB时,仍能准确地估计目标的方位.

Correlation peak detection and azimuth estimation based on empirical mode decomposition

When there is a low SNR,correlation peaks cannot be detected by the traditional cross-correlation method,so a target's azimuth cannot be estimated effectively.However,the empirical mode decomposition(EMD) method,which can emphasize the local instantaneous characteristics of the original signal,can decompose the original signal into several intrinsic mode functions(IMF).In this paper cross-correlation functions were decomposed with EMD method,then cross-correlation functions were reconstructed by suitably selecting decomposed IMFs.The correlation peak in the reconstructed waveform is obvious,which can substantially improve the detection ability.Hence,combining EMD with cross-correlation azimuth estimation using two hydrophones improved the estimation of time-delay and target's azimuth even with a low SNR.Simulation results proved that,compared the reconstructed cross-correlation function with the original one,the correlation peak is raised by 4.9 dB than the highest false peak,therefore achieving better time-delay estimate.Azimuth estimation results showed that the improved method can estimate a target's azimuth correctly when the SNR is-14 dB.