Deep convolution stack for waveform in underwater acoustic target recognition

Shengzhao Tian,Jingfa Liu,Hang Wang,Duanbing Chen

doi:10.1038/s41598-021-88799-z

Abstract

In underwater acoustic target recognition, deep learning methods have been proved to be effective on recognizing original signal waveform. Previous methods often utilize large convolutional kernels to extract features at the beginning of neural networks. It leads to a lack of depth and structural imbalance of networks. The power of nonlinear transformation brought by deep network has not been fully utilized. Deep convolution stack is a kind of network frame with flexible and balanced structure and it has not been explored well in underwater acoustic target recognition, even though such frame has been proven to be effective in other deep learning fields. In this paper, a multiscale residual unit (MSRU) is proposed to construct deep convolution stack network. Based on MSRU, a multiscale residual deep neural network (MSRDN) is presented to classify underwater acoustic target. Dataset acquired in a real-world scenario is used to verify the proposed unit and model. By adding MSRU into Generative Adversarial Networks, the validity of MSRU is proved. Finally, MSRDN achieves the best recognition accuracy of 83.15%, improved by 6.99% from the structure related networks which take the original signal waveform as input and 4.48% from the networks which take the time-frequency representation as input.

Highlights

Representation of underwater target acoustic signals in an unsupervised manner
Yue et al.[27] compared Deep Belief Network (DBN) to Convolutional Neural Network (CNN) using spectrogram as the input of networks, and the results showed that deep learning methods can achieve higher recognition accuracy
multiscale residual deep neural network (MSRDN) achieves the best recognition accuracy of 83.15%, improved by 6.99% from the structure related networks which take the original signal waveform as input and 4.48% from the networks which take the time-frequency representation as input

Summary

Introduction

Representation of underwater target acoustic signals in an unsupervised manner. The recognition performance was improved greatly compared with the traditional methods. Yue et al.[27] compared DBN to Convolutional Neural Network (CNN) using spectrogram as the input of networks, and the results showed that deep learning methods can achieve higher recognition accuracy. Yang et al.[19] designed a bank of multiscale deep convolution filters to decompose raw time domain signal into signals with different frequency components and made an improvement by refining the fusion and classification layers of depth characteristics. It achieved a classification accuracy of 81.96%. In underwater acoustic target recognition, Jin et al.[43] utilized GAN to extend the dataset by generating LOFAR spectrogram, and improved the performance of classification. The general deep network structure may not be suitable for underwater acoustic target recognition

Methods

Results

Discussion

Conclusion