【作者】 翟铎； 李风华；

【Author】 ZHAI Duo;LI Feng-hua;The Institute of Acoustics of the Chinese Academy of Sciences;University of Chinese Academy of Science;

【机构】 中国科学院声学研究所； 中国科学院大学；

【摘要】 浅海环境中当单频点源距离水平阵足够远时,各阶模态可以等效为入射方向不同的平面波,接收到的信号则是多个平面波的线性叠加。利用这个特点,从理想的空间谱中可以提取出各阶模态的水平波束及强度,然而常规波束形成(CBF)想要达到这个目的,受限于阵列孔径,为了克服孔径的限制,引入了变分贝叶斯。变分贝叶斯属于压缩感知的范畴是分层贝叶斯模型与变分推断的结合,接收信号的先验分布假设成Student-t先验,信号矢量的后验概率的期望即是估计到的空间谱。通过仿真实验证明了从变分贝叶斯估计到的空间谱中可以提取各阶模态的水平波束及相对强度并且与CBF进行了比较,此外给出了空间谱随声源深度及频率变化的理论值和估计值。更多还原

【Abstract】 In shallow water environment,when the point source at signal frequency bin is far enough from the horizontal array,the modes can be equivalent to plane waves from different directions and the received signals are the linear superposition of multiple plane waves.Using this feature,the horizontal beam and intensity of each mode can be extracted from the ideal spatial spectrum.However,conventional beamforming is limited by the array aperture.In order to overcome the limitation of aperture,variational Bayesian method is introduced.Variational Bayesian is the combination of hierarchical Bayesian model and variational inference,which belongs to the compressive sensing.The prior distribution of the received signals is assumed to be Student-t distribution.The expectation of the posterior probability of signal vector is the estimated spatial spectrum.The simulation results show that the horizontal beam and relative intensity of each mode can be extracted from the spatial spectrum estimated by the variational Bayesian method and compared with the CBF.In addition,the theoretical and estimated values of the spatial spectrum varying with the depth and frequency of the sound source are given.更多还原