一种高精度的神经网络波达方向估计算法

A High Precision Neural Network Direction of Arrival Estimation Algorithm

基于区间划分的神经网络DOA (波达方向)估计算法通过建立多个网络结构来进行波达方向的估计,但是对一些特定角度的信号,识别的精确度不高。为了提高神经网络算法的精确度和适用性,本文对该算法进行了改进:首先利用波束形成技术,调整天线阵元的权值,将天线阵的波束集中在每个区间所对应的角度;在区间划分上,采用边缘重叠的区间划分方式替代原有的均匀无重叠方式。相比于之前的神经网络算法,该算法的识别精确度更高,并且对于任意角度的信号都适用,最后通过仿真验证了该算法的正确性与有效性。

The neural network DOA (Direction of Arrival) estimation algorithm based on interval division establishes multiple network structures to estimate the direction of arrival, but the accuracy of recognition is not high for some signals at specific angles. In order to improve the accuracy and applicability of the neural network algorithm, this article has improved the algorithm: firstly, the beamforming technology is used to adjust the weight of the antenna array element, and the beam of the antenna array is concentrated at the angle corresponding to each interval;in terms of interval division, the interval division method with edge overlap is used to replace the original uniform and non-overlap method. Compared with the previous neural network algorithm, this algorithm has higher recognition accuracy and is applicable to signals of any angle. Finally, the correctness and effectiveness of the algorithm are verified by simulation.