Time-Domain Anti-Interference Method for Ship Radiated Noise Signal

Abstract Ship radiated noise signal is one of the important ways to detect and identify ships, emission interference shield its own a common countermeasure. In this paper, we try use idea enhancement enhance ship with extremely low signal-to-noise ratio, so as achieve anti-explosive interference. We propose deep learning model by mask in temporal domain. Our approach an encoder–decoder structure U-net. U-net consists 1d-conv skip connection. order improve ability model, directly connect series. model’s time series information. The Transformer attention mechanism adopted make have learn combine Loss function for scale-invariant source-to-noise ratio mean squared error time-domain. Finally, actual collected data conduct experiments. It verified that our algorithm can effectively 2 dB under − 20 25 dB.