基于神经网络在矢量水听器中的波达方向角的应用研究

发布时间：2018-03-05 12:53

本文选题：波达方向角　切入点：矢量水听器　出处：《中北大学》2017年硕士论文　论文类型：学位论文

【摘要】：与标量水听器作对比,矢量水听器具有体积小、一致性好等优点,其制作流程和相关技术日渐成熟,因此关于矢量水听器的研究是未来研究趋势之一;而人工神经网络作为机器学习的代表,是一种自学习、自适应和并行分布的信息处理网络结构。所以,运用人工神经网络来解决矢量水听器的波达方向角估计问题不失为一种良好的选择。本文系统地研究了矢量水听器的信号处理,通过模型假设、特征提取、网络构建和仿真实验,有效地验证了本文所提方法在算法复杂度和精确度等方面比其他方法优秀。论文主要成果有:1.第三章介绍了基于粒子群算法优化BP神经网络的矢量水听器的DOA估计。在文中详尽的叙述了适合矢量水听器的信号特征提取方法,将提取的信号特征作为神经网络的输入,网络模型则选用对数据具有非线性拟合能力的PSO-BP网络。通过仿真实验比较,说明该方法不仅特征提取方法简便,精度和算法运行时间都优于其他文献算法。2.第四章阐述了基于果蝇算法优化广义回归神经网络的矢量水听器的DOA估计。首先介绍了果蝇优化算法和广义回归神经网络的工作原理,然后探讨了果蝇算法是如何优化广义回归神经网络。仿真实验结果表明,本章所构建的网络模型与其他文献方法相比,结构简洁,来波方向精度更高,运行时间更短。关于矢量水听器的研究是一个复杂的问题,未来仍需大量的深入研究工作,真挚地希望本文能对矢量阵列信号处理技术的研究和发展有所帮助。
[Abstract]:Compared with the scalar hydrophone, the vector hydrophone has the advantages of small size, good consistency and so on. The manufacturing process and related technology of the vector hydrophone are becoming more and more mature, so the research on the vector hydrophone is one of the future research trends. As the representative of machine learning, artificial neural network is a self-learning, adaptive and parallel information processing network structure. It is a good choice to use artificial neural network to solve the problem of DOA estimation of vector hydrophone. In this paper, the signal processing of vector hydrophone is studied systematically. Network construction and simulation experiments, It is proved that the proposed method is superior to other methods in complexity and accuracy. The main achievements of this paper are: 1. Chapter 3 introduces the DOA estimation of vector hydrophone based on particle swarm optimization (PSO) and BP neural network. In this paper, the method of signal feature extraction suitable for vector hydrophone is described in detail. The extracted signal feature is used as the input of the neural network, and the PSO-BP network with nonlinear fitting ability for the data is selected in the network model. The simulation results show that this method is not only simple, but also convenient for feature extraction. In chapter 2. 4th, the DOA estimation of vector hydrophone based on generalized regression neural network based on Drosophila algorithm optimization is discussed. Firstly, the optimal algorithm and generalized regression neural network of Drosophila are introduced. How the network works, Then it discusses how the algorithm optimizes the generalized regression neural network. The simulation results show that the network model constructed in this chapter is simpler in structure and more accurate in the direction of wave than other methods in literature. The research on vector hydrophone is a complex problem, and a lot of further research work is needed in the future. It is sincerely hoped that this paper will be helpful to the research and development of vector array signal processing technology.
【学位授予单位】：中北大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP183;TB565.1

【相似文献】