针对现有测向系统多信号适应能力弱、测向精度低的问题，提出一种基于数据驱动的高精度阵列测向新方法。该方法提取单信号入射时的输入特征向量，基于卷积神经网络构建单信号测向网络框架。利用信号的独立性，将多信号测向问题转化为单信号测向问题，在单信号训练网络的基础上实现多信号来波方向估计。仿真实验与理论分析结果表明，该方法有效减少了输入特征维数和网络训练样本数目，在多信号同时入射及阵列互耦效应条件下均获得了高精度的到达方向(Direction of Arrival，DOA)估计的测向结果。
High Precision Array Direction Finding Method Based on Data Driven
Aiming at the problems of weak multi-signal adaptability and low direction finding estimatesaccuracy of existing array di-rection finding systems,a new data-driven intelligent array direction finding method is proposed.Firstly,the input features under the condition of uniform linear array of single-signal incidence are extracted,a single-signal direction finding network framework is construc-ted based on convolutional neural network,and then the multi-signal direction finding problem is transformed into a single-signal direc-tion finding problem by using the independence of the signal,and the multi-signal direction estimation is realized on the basis of the sin-gle-signal training network.Simulation experiments and theoretical analysis results show that the proposed method effectively reduces the input feature dimension and the number of network training samples,and it obtains high Direction of Arrival(DOA)estimates accuracy under multi-signal and mutual coupling effects.
convolutional neural networkdata drivenarray direction findingmutual coupling effects
国家科技期刊平台
NSTL
万方数据
