Deep learning method for distinguishing shallow sea surface and underwater target based on vertical array

The invention discloses a deep learning method for distinguishing shallow sea surface and underwater targets based on a vertical array, and belongs to the field of underwater sound signal processing. In order to solve the problem that in the prior art, feature distribution of water surface and underwater targets is overlapped in a feature space, the invention provides a deep learning method for distinguishing shallow sea water surface and underwater targets based on a vertical array, and aims to solve the problem that feature distribution of the water surface and underwater targets is highly similar due to similar physical characteristics and environmental parameters. According to the technical scheme, the method comprises the following steps: collecting hydrological data for training corresponding to a shallow sea surface underwater target for training; obtaining a coherence spectrum data set of the corresponding sound field according to the hydrological data for training; performing feature extraction and n