Underwater acoustic source localization using generalized regression neural network.

Area of Science:

• Ocean Acoustics
• Machine Learning
• Signal Processing

Background:

• Underwater source localization is crucial for marine applications.
• Traditional methods often require accurate environmental models.
• Existing machine learning approaches may have limitations.

Purpose of the Study:

• Introduce a novel machine learning approach for underwater source localization.
• Evaluate the performance of Generalized Regression Neural Network (GRNN) for this task.
• Compare GRNN with other methods like Feed-Forward Neural Network (FNN) and Matched Field Processing (MFP).

Main Methods:

• Source localization framed as a supervised learning regression problem.
• Utilizing Generalized Regression Neural Network (GRNN) with normalized sample covariance matrix (SCM) as input.
• Optimizing GRNN's spread factor using cross-validation.
• Experimental validation using SWellEx-96 vertical array data.

Main Results:

• GRNN demonstrated superior localization performance compared to FNN and MFP.
• The proposed GRNN method is data-driven, requiring minimal a priori information.
• Satisfactory localization accuracy was achieved.

Conclusions:

• GRNN offers a robust and effective alternative for underwater source localization.
• This method is particularly advantageous when environmental information is scarce or uncertain.
• The data-driven nature of GRNN simplifies the localization process.