Robust Dual Target Range Estimation Method of Vertical Array Using Residual Convolution Neural Network
Authors:
- 1 November 2022
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 742-747
Abstract
In the actual marine environment, it is often necessary to locate multiple underwater targets at the same time. Aiming at the problem of simultaneous localization of multiple sound sources, this paper studies the double target range estimation scene in the same depth and uses normal wave model to generate the underwater acoustic data of dual targets. The range estimation of dual targets in the same sea area and different sea areas is carried out, and two frequency combinations of dual targets, namely 280/385Hz and 49/388Hz, are taken as examples to analyze the impact of different frequencies on the range estimation results. The network structure adopts the convolution neural network (CNN) with residual connection. The simulation results show that the residual CNN model can effectively realize the range estimation of dual targets in the same sea area and different sea areas, and the greater the frequency difference, the more accurate the range estimation results of the dual targets. Therefore, all results show that the proposed method has excellent performance for the dual target range estimation.Keywords
Funding Information
- National Natural Science Foundation of China (11974286)
This publication has 10 references indexed in Scilit:
- Model-based convolutional neural network approach to underwater source-range estimationThe Journal of the Acoustical Society of America, 2021
- Seabed and range estimation of impulsive time series using a convolutional neural networkThe Journal of the Acoustical Society of America, 2020
- Deep-learning source localization using multi-frequency magnitude-only dataThe Journal of the Acoustical Society of America, 2019
- Source Ranging Using Ensemble Convolutional Networks in the Direct Zone of Deep WaterChinese Physics Letters, 2019
- Source localization using deep neural networks in a shallow water environmentThe Journal of the Acoustical Society of America, 2018
- Sound Source Localization in a Multipath Environment Using Convolutional Neural Networks2018
- Underwater acoustic source localization using generalized regression neural networkThe Journal of the Acoustical Society of America, 2018
- Ship localization in Santa Barbara Channel using machine learning classifiersThe Journal of the Acoustical Society of America, 2017
- Deep Residual Learning for Image Recognition2016
- Neocognitron: A neural network model for a mechanism of visual pattern recognitionIEEE Transactions on Systems, Man, and Cybernetics, 1983