Research on underwater target recognition based on auditory EEG signal features and deep learning

Target recognition is a key technical link in hydroacoustic detection, which has a broad application prospect in the fields of marine safety and resource exploration. In recent years, artificial involvement of underwater target recognition methods based on analysis of line spectra, auditory spectra and other characteristics are broadly utilized in actual engineering applications, with their unique characteristics. Meanwhile, the learning tasks of human-computer interaction have been widely used, for machine learning and deep learning are also developing rapidly. Therefore, in this paper, auditory EEG signals under the excitation of underwater targets' signals are used to carry out recognition research by combining human brain perception, artificial analysis, SVM and ResNet with atrous convolution. The results show that the recognition rate of four types of ship radiated noise can reach 94.35% using the ResNet with atrous convolution, which can effectively classify and recognize underwater targets.