A feedforward neural network for source range and ocean seabed classification using time-domain features

Acoustic source ranging in an uncertain ocean environment is a complicated problem, though classification and regression-based machine learning algorithms show promise. A feedforward neural network (FNN) has been trained to do either classification or regression on both the source-receiver range and ocean seabed type using extracted time-domain features. Pressure time-series are generated to simulate signals received at different ranges in three different ocean environments, representing sandy, muddy, and mixed sediment seabeds. Four features are extracted from these waveforms: peak level, integrated level, signal length, and decay time. These four features are used to train FNN for both classification and regression of range and environment type, and the results are compared to a network trained on the time waveforms. Even for small amounts of training data, the pressure time-series provide a higher accuracy than the extracted features. These results lay a foundation for comparisons to the more computationally expensive convolutional neural networks.