Underwater Acoustic Target Recognition Based on Multi-scale Residuals and Dual Attention Mechanism

Aiming at the problems that traditional ship underwater acoustic target recognition methods are difficult to achieve high recognition accuracy and have low robustness in complex underwater environments, this paper proposes a target recognition model that integrates multi-scale residual convolution and dual attention mechanism(MSR-DAN). A multi-scale residual convolution module is designed to capture the local and global features of acoustic signals. Combined with channel attention and spatial attention, the adaptive allocation of feature weights is achieved to enhance the expression of the characteristic information of ship radiated noise. The experimental results show that this method can achieve a recognition accuracy rate of 97.86% on the public ship dataset. Under the condition of low signal-to-noise ratio, it can also obtain an accuracy rate superior to other models.