Torpedo artificial jamming countermeasure method based on attention gate wasserstein generative adversarial networks

This research investigates the challenges posed by artificial jamming in torpedo target detection. The study provides a concise exploration of jamming scenario applicable to both fuze and guidance system. Drawing insights from the successful application of Generative Adversarial Networks (GANs) in enhancing speech signals, we leverage Wasserstein GAN while incorporating the unique features of underwater acoustic signals. Concerning the analysis of the jamming scenario, we have designed a specific training set for the anti-jamming purpose in torpedo detection. We have proposed Attention Gate Wasserstein Generative Adversarial Networks. We incorporate an L2 loss term into the loss function to enforce a one-to-one mapping between the input and output. Furthermore, given that the soft attention mechanism promotes determinism and enhances model performance, we integrate attention gates into the generator to promote determinism and enhance model performance. Through comprehensive simulation analysis, we demonstrate the efficacy of the trained AGWGAN generator in effectively mitigating the impact of the three types of artificial jamming signals.