Underwater acoustic target recognition method based on a joint neural network.

Area of Science:

• Marine acoustics
• Artificial intelligence
• Signal processing

Background:

• Underwater acoustic target recognition is crucial for marine surveillance and research.
• Existing artificial neural network methods face challenges in achieving high recognition accuracy.
• The integration of deep learning models offers potential for improved performance.

Purpose of the Study:

• To develop and evaluate a novel hybrid neural network for underwater acoustic target recognition.
• To enhance the accuracy and classification capabilities compared to single network models.
• To explore the application of deep learning in classifying underwater acoustic signals.

Main Methods:

• A novel network framework integrating a one-dimensional convolutional neural network (1D CNN) and a long short-term memory (LSTM) network was designed.
• Ship acoustic data were utilized as input for training and testing the proposed model.
• Performance was evaluated through visual analysis and comparison with single neural network approaches.

Main Results:

• The integrated 1D CNN-LSTM network demonstrated effective recognition and classification of underwater acoustic targets.
• The proposed hybrid model achieved considerably higher recognition accuracy than single neural network models.
• Visual analysis confirmed the robustness and effectiveness of the new recognition method.

Conclusions:

• The hybrid 1D CNN-LSTM network offers a significant advancement in underwater acoustic target recognition.
• This deep learning approach provides a promising new direction for marine acoustic signal analysis.
• The findings suggest broad applicability of this integrated network in underwater surveillance and research.