HHT和SVM在4-UCA信源数估计中的应用

Applicationof HHT and SVM in the Signal Source Number Estimation of four uniform circular arrays

四均匀圆阵(4-UCA)广泛应用在阵列信号处理的来波方向(DOA)估计中,而正确估计出信号源个数有利于提高DOA估计的性能。基于4-UCA,利用HHT自适应分解信号的优点,通过求取阵列间希尔伯特相位相关系数矩阵特征值来对信号源提取特征,并把特征值输入SVM训练分类器,从而将信号源个数估计的问题转换为模式识别中的分类问题来解决。通过采集实际数据实验,结果显示,正定情况下,该方法对信号源个数有较好的识别效果,同时检测性能具有鲁棒性。

Four array uniform circular array is widely used in DOA estimation of array signal processing.The number of signal sources can be estimated is the key to high resolution DOA estimation.In this paper,a method combining Hilbert Huang Transform and Support Vector Machine for estimating the number ofsignal source is proposed in four uniform circular arrays. The array signals are decomposed by EMD at first, and the intrinsic mode functions are obtained.Then the Hilbert instan-taneous phase of intrinsic mode functions are obtained. Inaddition,these sets of phase are adopted to construct Hilbert instan-taneous phasecorrelation coefficient matrix of the corresponding IMF components between each array. Finally,the eigenvalues of the correlation coefficient matrix ,which axe input SVM to train SVM network,are used to estimatethe number of signal sources. Real-Iife experiment results show that the proposed method has good performance of signal source estimation and the performance of estimation is robust.Key words：4 uniform circular arrays ;source number estimation;Hilbert Huang Transfonn;correlation coefficient;featuie extrac-