Abstract: In array signals estimation，ESPRIT algorithm is a common algorithm of high speed and precision，but it don’t estimate mixed signal（including coherent and non-coherent synchronously） that its coherent is unknown in low SNR.Combining the excellent performance on MMUSIC and ESPRIT algorithm，a new improved ESPRIT method is proposed for DOA estimation which includes coherent and non-coherent.The method makes use of correlation and cross correlation information of covariance matrix to construct the new matrix which includes signal direction data.DOA is gottten from the decomposition of above matrix.Simulation show this method has advantage and dependability.

Key words: direction-of-arrival, mixed signal, covariance matrix, cross-correlation

摘要： 在目前信号波达方向（Direction-Of-Arrival，DOA）估计中，常规ESPRIT算法是一种速度快、精度高的常用算法，但对于低信噪比下混合信号（同时含有相干与非相干信号），常规ESPRIT算法难以估计出它们的DOA。结合解相干MUSIC和常规ESPRIT算法的优点，提出了一种新的估计相干与非相干信源的ESPRIT方法，新方法充分利用数据协方差矩阵的自相关和互相关信息来重构含有信号方位数据的新矩阵，再从它的特征值中解得信号的到达角。计算机仿真结果验证该方法在混合信号估计中的优越性和可靠性。

关键词: 波达方向, 混合信源, 协方差矩阵, 互相关