摘要
针对传统神经网络波达方位(DOA)估计算法抗噪性能差、精度不高以及泛化性弱等问题,提出基于局部加权长短时记忆神经网络的DOA估计算法。该算法对长短时记忆(LSTM)进行局部加权回归,将单阵元与全阵元接收信号功率之比作为网络的输入向量,根据分类结果选取对应区间数据进行训练,建立网络学习特征与DOA估计之间的非线性映射关系。该算法应用于无人机方向定位的实验表明,与传统神经网络的DOA算法相比,该算法具有更好的准确性、抗噪性和泛化性。
Aiming at the poor performance in anti-noise,low precision and weak generalization of traditional neural network DOA estimation algorithms,a DOA estimation algorithm based on local weighted long short term memory (LWLSTM) neural network is proposed. In this algorithm,the local weighted regression of long short-term memory (LSTM) is carried out,and the ratio of the received signal power of single array element to that of full array element is taken as the input vector of the network. According to the classification results,the corresponding interval data are selected for training,and the non-linear mapping relationship between the network learning characteristics and DOA estimation is established. The experimental results in the orientation of UAV show that the proposed algorithm has better accuracy,good in anti-noise and generalization comparing with the traditional neural network DOA algorithm.
作者
郭业才
施钰鲲
GUO Yecai;SHI Yukun(College of Electronic and Information Engineering,Nanjing University of Information Science and Technology,Nanjing 210044,China;Jiangsu Collaborative Innovation Center on Atmospheric Environment and Equipment Technology(CICAEET),Nanjing 210044,China;Bingjiang College,Nanjing University of Information Science and Technology,Wuxi 214105,Jiangsu,China)
出处
《实验室研究与探索》
CAS
北大核心
2020年第10期113-117,共5页
Research and Exploration In Laboratory
基金
国家自然科学基金项目(61673222)
南京信息工程大学滨江学院科研与教研项目(2019bjynk002,JGZDA201902)。
关键词
局部加权
长短时记忆网络
最大似然准则
无人机
波达方向
local weighted
long short term memory
maximum likelihood criterion
unmanned aerial vehicle(UAV)
direction of arrival(DOA)
