基于深度学习的人工智能在语音识别技术取得了突破性的进展，为被动声呐目标分类提供了新思路。该文提出一种将一维卷积神经网络与长短时记忆网络融合的深度学习分类模型（Conv-LSTM），提取了被动声呐目标听觉感知特征——梅尔频率倒谱系数（MFCC），并将特征输入模型提取深度特征进行目标分类。试验结果表明，该模型相较卷积神经网络和长短时记忆网络具有更高的识别率。
Artificial intelligence speech recognition technology based on deep learning has made a breakthrough in speech reconition technology, which provides new ideas for passive sonar target classification. In this paper, we propose a deep learning classification model(Conv-LSTM) that fuses a one-dimensional convolutional neural network with a long-short term memory network. We extract auditory sensing feature of a passive sonar target, the mel frequency cepstrum coefficient(mfcc), and the feature input model to extract deep features for classification passive sonar target. The experimental results show that Conv-LSTM model has higher recognition rate than convolutional neural network and long-short term memory network.
2020,42(10): 129-133 收稿日期：2020-02-20
DOI：10.3404/j.issn.1672-7649.2020.10.025
分类号：TB56
作者简介：杨路飞(1995-)，男，硕士研究生，研究方向为水声信号处理
参考文献：
[1] REMOTE S. Reports on remote sensing findings from Harbin Engineering University provide new insights (underwater acoustic target recognition: a combination of multi-dimensional fusion features and modified deep neural network)[J]. Computers, Networks & Communications, 2019.
[2] 陈晓伟, 杨明莉. 水下探测目标识别的BP网络模型的设计与实现[J]. 舰船科学技术, 2017, 39(2): 118-120
CHEN Xiaowei, YANG Mingli. The design and realization of BP network model for underwater detection target recognition[J]. Ship Science and Technology, 2017, 39(2): 118-120
[3] 王念滨, 何鸣, 王红滨, 等. 适用于水下目标识别的快速降维卷积模型[J]. 哈尔滨工程大学学报, 2019, 40(7): 1327-1333
[4] KE Xiaoquan, YUAN Fei, CHENG En. Underwater acoustic target recognition based on supervised feature-separation algorithm.[J]. Sensors (Basel, Switzerland), 2018, 18(12)
[5] HU Gang, WANG Kejun, PENG Yuan, et al. Deep learning methods for underwater target feature extraction and recognition.[J]. Computational intelligence and neuroscience, 2018: 2018
[6] YANG Honghui, SHEN Sheng, YAO Xiaohui, et al. Competitive deep-belief networks for underwater acoustic target recognition.[J]. Sensors (Basel, Switzerland), 2018, 18(4)
[7] 陈冬, 李钢虎, 赵亚楠. 基于MVDR的MFCC方法在水下目标识别中的应用[J]. 声学与电子工程, 2013(3): 23-25
[8] 徐新洲, 罗昕炜, 方世良, 等. 基于听觉感知机理的水下目标识别研究进展[J]. 声学技术, 2013, 32(2): 151-158
[9] 张大伟, 章新华. 基于听觉感知与卷积神经网络识别舰船目标[J]. 声学技术. 2015. 06.
[10] LECUN Y, BENGIO Y, HINTON G. Deep learning[J]. Nature, 2015, 521(7553): 436-444
[11] HINTON G E, OSINDERO S, TEH Y W. A fast learning algorithm for deep belief nets.[J]. Neural Computation, 2006, 18(7): 1527-1554
[12] HINTON G E, SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks.[J]. Science, 2006, 313(5786): 504-7
[13] KRIZHEVSKY, Alex, ILYA Sutskever, and GEOFFREY E. Hinton. Imagenet classification with deep convolutional neural networks. Advances in neural information processing systems. 2012.
[14] HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural computation, 1997, 9(8): 1735-1780
[15] JOHNSON J, KARPATHY A, FEI-FEI L. DenseCap: Fully Convolutional Localization Networks for Dense Captioning[J]. 2015.
[16] 韩云飞, 张太红, 白涛. 融合Mel系数和kNN的语音端点检测[J]. 信息技术, 2017(3): 37-41
[17] 邵央, 刘丙哲, 李宗葛. 基于MFCC和加权矢量量化的说话人识别系统[J]. 计算机工程与应用, 2002(5): 127-128
[18] 苏赋, 吕沁, 罗仁泽. 基于深度学习的图像分类研究综述[J]. 电信科学, 2019, 35(11): 58-74
[19] CHUNG J, GULCEHRE C, CHO K H, et al. Empirical evaluation of gated recurrent neural networks on sequence modeling[J]. arXiv preprint arXiv: 1412.3555, 2014
