应用SAPSO-BP神经网络的DOA估计方法  

DOA Estimation Method Based on SAPSO-BP Neural Network

作　　者：姚建丽 胡红萍[1] 白艳萍[1] 王建中[1] YAO Jianli, HU Hongping, BAI Yanping, WANG Jianzhong(School of Science,North University of China ,Taiyuan 030051 ,China)

机构地区：[1]中北大学理学院,太原030051

出　　处：《重庆理工大学学报（自然科学）》2018年第5期183-188,共6页Journal of Chongqing University of Technology：Natural Science

基　　金：国家自然科学基金资助项目(61774137);山西省自然科学基金资助项目(201701D22111439;201701D221121)

摘　　要：声源方向估计是声纳、雷达、无线电发射机跟踪和移动通信中的基本问题之一。对矢量水听器声源的波达方向角(DOA)进行研究。传统的BP神经网络容易陷入局部最优,虽然PSO优化的BP神经网络在一定程度上改善了这个缺点,但仍容易早熟收敛,造成搜索精度的降低。为此,提出了一种模拟退火粒子群算法,并利用其优化BP神经网络,改进矢量水听器声源的波达方向角(DOA)估计的性能。仿真实验结果表明:模拟退火粒子群算法优化的BP神经网络具有更好的泛化能力,提高了DOA的估计精度。Sound source direction estimation is one of the basic issues in sonar, radar, radio transmitter tracking and mobile communications. In this paper,the DOA of vector hydrophone source is studied. The traditional BP neural network is easy to fall into the local optimum. Although the PSOoptimized BP neural network improves this disadvantage to some extent,it is still easy to converge prematurely,resulting in the reduction of search accuracy. In this paper,a simulated annealing particle swarm optimization algorithm is proposed,and uses its optimized BP neural network to improve the DOA estimation performance of the vector hydrophone source. Simulation results show that the BP neural network optimized by simulated annealing particle swarm optimization algorithm has better generalization ability and improves the DOA estimation accuracy.

关 键 词：粒子群算法 模拟退火算法 矢量水听器 BP神经网络 波达方向角 

分 类 号：TN911.7[电子电信—通信与信息系统]