A Deep Convolutional Neural Network Inspired by Auditory Perception for Underwater Acoustic Target Recognition

Abstract

Underwater acoustic target recognition (UATR) using ship-radiated noise faces big challenges due to the complex marine environment. In this paper, inspired by neural mechanisms of auditory perception, a new end-to-end deep neural network named auditory perception inspired Deep Convolutional Neural Network (ADCNN) is proposed for UATR. In the ADCNN model, inspired by the frequency component perception neural mechanism, a bank of multi-scale deep convolution filters are designed to decompose raw time domain signal into signals with different frequency components. Inspired by the plasticity neural mechanism, the parameters of the deep convolution filters are initialized randomly, and the is n learned and optimized for UATR. The n, max-pooling layers and fully connected layers extract features from each decomposed signal. Finally, in fusion layers, features from each decomposed signal are merged and deep feature representations are extracted to classify underwater acoustic targets. The ADCNN model simulates the deep acoustic information processing structure of the auditory system. Experimental results show that the proposed model can decompose, model and classify ship-radiated noise signals efficiently. It achieves a classification accuracy of 81.96%, which is the highest in the contrast experiments. The experimental results show that auditory perception inspired deep learning method has encouraging potential to improve the classification performance of UATR.

Underwater Target Signal Classification Using the Hybrid Routing Neural Network

“…In Figure 9 , the used network is compared with DNN [ 13 ], ADCNN [ 15 ], DepthCNN [ 16 ], CDBN [ 17 ], and CRNN [ 18 ]. HRNet is the hybrid routing network method.…”

“…An end-to-end learning network, called the auditory perception inspired deep CNN (ADCNN), is an efficient network architecture in signal feature extraction, and the method transform signals from the temporal domain to the frequency domain. The deep representation of raw signal can be separated; this method achieves satisfactory performances in underwater acoustic target classification [ 15 ]. The original signal data are input into the depthwise separable CNN (DSCNN) in the temporal domain.…”

Passive Detection of Ship-Radiated Acoustic Signal Using Coherent Integration of Cross-Power Spectrum with Doppler and Time Delay Compensations

“…Compared with other methods, it is shown that the proposed method is capable of providing sufficient processing gain and effectively enhancing the narrowband discrete components in a ship-radiated acoustic signal. This is also of great significance for target recognition [21,22] and classification [23].…”

Design and Performance Evaluation of a Deep Neural Network for Spectrum Recognition of Underwater Targets

“…In the recent years, as the most popular deep learning model, the deep neural network (DNN) has attracted the interest of scholars in the field of UATR [16]. Yang et al combined the auditory perception principle and convolutional neural network (CNN) to propose an auditory perception-inspired deep convolutional neural network (ADCNN), which used a CNN to extract features of different frequency components from signals and merged them at the fusion layer to achieve the classification of acoustic targets [17]. Choi et al used the absolute values of matrix elements in the cross-spectrum density matrix (CSDM) to generate two additional matrices as input data and used them for training the CNN model [18].…”