This paper addresses the direction of arrival (DOA) estimation via an acoustic vector sensor (AVS) array in the presence of non-orthogonal factors. To mitigate the estimation bias, which is caused by non-orthogonal factors, a novel DOA estimator is proposed by combining the iterative sparse maximum likelihood-based and maximum a posteriori (ISML-MAP) approaches. First, the two non-orthogonal AVS array models are formulated by introducing a perturbation parameter, based on which the DOA estimation bias is quantified for a single-source scenario. The results show that the non-orthogonal factor has a greater influence on the DOA estimation performance when the velocity sensor located in x-axis is selected as the reference sensor compared to the non-orthogonal AVS array model with the velocity sensor located in y-axis selected as the reference sensor. Then, the DOA of the acoustic source and the non-orthogonal deviation matrix (or angle deviation) are jointly estimated iteratively. In each iteration, three matrix rotation approaches are presented to determine the non-orthogonal deviation matrix. Simulation results demonstrated that the proposed methods achieve better DOA estimation performance than the existing methods for the non-orthogonal AVS array.

本文在存在非正交因素的情况下，通过声矢量传感器（AVS）阵列解决了到达方向（DOA）估计问题。为了减轻由非正交因素引起的估计偏差，通过结合基于稀疏最大似然和最大后验（ISML-MAP）的方法，提出了一种新颖的DOA估计器。首先，通过引入扰动参数来制定两个非正交的AVS阵列模型，基于此参数，可以量化单源场景的DOA估计偏差。结果表明，非正交的因子对DOA估计性能有较大的影响，当位于所述速度传感器X轴被选择为与具有位于所述速度传感器非正交AVS阵列模型中的基准传感器y轴被选作参考传感器。然后，迭代地联合估计声源的DOA和非正交偏差矩阵（或角度偏差）。在每次迭代中，提出了三种矩阵旋转方法来确定非正交偏差矩阵。仿真结果表明，与现有的非正交AVS阵列方法相比，该方法具有更好的DOA估计性能。