Underwater signal prediction and parameter estimation using artificial neural networks

A model of sound propagation in underwater layered media (UWLM) accounting for attenuation effects is employed to test artificial neural networks' ability in signal prediction and parameter estimation. Two fully interconnected feed-forward multilayered neural networks with necessary layers trained by back-propagation supervised learning algorithm using the min-max amplitude ranges of the output signals of UWLM are designed and evaluated. These are based on synthetic data, to estimate the parameters of the media including attenuation factors, reflection coefficients, travel times and decay values. Based on experiments estimating the parameters of the media and predicting its output signal, the networks produce results very close to those of the original assumed media structure The results suggest that the proposed networks can supplement, or replace conventional techniques for parameter estimation and output prediction in system identification. The method presented also offers advantages in speed and efficiency over existing estimates techniques